Thompson's 1997 working paper — independent of his 2007 paper — describes Argote, Beckman & Epple (1990) as adding labor turnover to Rapping's (1965) specification, not as a rival re-analysis

economicslearning-curveorganizational-forgettingliberty-shipsargoterappingpeter-thompsonsource-lineage

A second, independent Peter Thompson paper — his 1997 working paper on Liberty-ship learning, not the 2007 paper on Liberty-ship forgetting already in the vault — places Argote, Beckman & Epple (1990) in a specific methodological lineage relative to Leonard Rapping's 1965 "Learning and World War II Production Functions." Thompson writes that "Rapping (1965) and Argote, Beckman and Epple (1990) used the number of authorized ways in each yard as a proxy for the capital stock" in the same yard-level production-function specification, and that "Argote, Beckman and Epple (1990) have added labor turnover rates to Rapping's analysis without undermining his key findings." Rapping's original specification is a Cobb-Douglas-type production function (ways, employment hours, cumulative output) pooled across fifteen Liberty-ship yards; Argote, Beckman & Epple's contribution, on this description, was to extend that specification with a labor-turnover term rather than to replace it or re-analyze the same case from scratch.

This corroborates, from a source independent of both Benkard's (2000) and Thompson's own 2007 citations, that Argote, Beckman & Epple's paper builds directly on Rapping's founding study — the root the whole Liberty-ship learning/forgetting cluster traces to. It does not supply the organizational-forgetting depreciation figure itself: Thompson's 1997 paper is about capital deepening and quality change at one yard (Calship), not about re-estimating forgetting, and quotes Argote, Beckman & Epple only for what they did, not what they found.