Choosing effective dates from multiple optima in Technology Forecasting using Data Envelopment Analysis (TFDEA)

• We demonstrate a computational issue raised in an earlier TFSC paper that occurs frequently.
• This issue results in different software programs generating different results.
• This paper provides a methodological fix for this computational issue.
• The methodological fix is demonstrated on previously published aerospace applications.
• This fix should be used in all future TFDEA applications using dynamic frontier years.

Technology Forecasting using Data Envelopment Analysis (TFDEA) provides an effective means to forecast technological capability over time without the burden of fixed a priori weighting schemes. However, there are situations where result reproduction can be a challenge as first pointed out in a previous Technological Forecasting and Social Change article [11]. When using a commonly used extension of TFDEA, there are circumstances where multiple optimal solutions can complicate analysis. This paper addresses this issue through extending the TFDEA model in a manner consistent with common Data Envelopment Analysis (DEA) techniques. The extension is then demonstrated using datasets from previous publications on fighter jet and commercial airplane technology where the issue of multiple optima has been observed. The results indicate that traditional TFDEA may generate varying forecasts depending on the software used, which can be dealt with by introducing a secondary objective function. Therefore, researchers should explicitly state which secondary objective function they are using for the TFDEA applications.