Khanna, Guler, Nerkar (2016) – Fail often, fail big, and fail fast? Learning from small
failures and R&D performance in the pharmaceutical industry
Summary
Under what conditions do firms learn from their failed innovation attempts (in Pharma)?
3 conditions are investigated: number of failures, importance of failures and timing of
failures (early/late)
2 dependent variables: output of R&D (no of patents) and quality of R&D outputs (citations
of patents)
Findings:
The higher the number of failures, the more important these failures to the firm and the
earlier these failures occur, the lower the amount of R&D outputs and the higher the quality
of these outputs
Why:
Quality increases because:
o Higher number of failures = firms gain a better sense of causal relationships
o Higher importance of failures = decision-makers will pay attention to these failures
o Earlier occurrence of failures = easier to pinpoint the factor causing the failure
Output decreases because:
o Learning from failures is a multi-level process. Patent decontinuation (=failure) is
decided at the firm level, however, idea generation happens at the individual
scientist level. When patents are discontinued, there is a time lag in feedback to the
individual level, causing the scientist to develop fewer ideas from failures
Findings in detail – Impact of failures on quality of R&D output
The quality of R&D output is positively influenced by 3 factors:
1) A high number of failures
2) A high importance (within the firm) of the failed projects
3) An early detection of failure (in the R&D process)
The more failures, the higher the quality of R&D outputs because:
Failure leads to feedback to make improvements
Failures initiates the search for the cause of failures
Failures are associated with deliberate learning process (inductively coming up with
innovative ideas)
Failure = rejection of hypotheses, so the next hypothesis can be supported
Failure stimulates distant search outside of knowledge corridors, resulting in more possible
solutions, from which the best one can be chosen
failures and R&D performance in the pharmaceutical industry
Summary
Under what conditions do firms learn from their failed innovation attempts (in Pharma)?
3 conditions are investigated: number of failures, importance of failures and timing of
failures (early/late)
2 dependent variables: output of R&D (no of patents) and quality of R&D outputs (citations
of patents)
Findings:
The higher the number of failures, the more important these failures to the firm and the
earlier these failures occur, the lower the amount of R&D outputs and the higher the quality
of these outputs
Why:
Quality increases because:
o Higher number of failures = firms gain a better sense of causal relationships
o Higher importance of failures = decision-makers will pay attention to these failures
o Earlier occurrence of failures = easier to pinpoint the factor causing the failure
Output decreases because:
o Learning from failures is a multi-level process. Patent decontinuation (=failure) is
decided at the firm level, however, idea generation happens at the individual
scientist level. When patents are discontinued, there is a time lag in feedback to the
individual level, causing the scientist to develop fewer ideas from failures
Findings in detail – Impact of failures on quality of R&D output
The quality of R&D output is positively influenced by 3 factors:
1) A high number of failures
2) A high importance (within the firm) of the failed projects
3) An early detection of failure (in the R&D process)
The more failures, the higher the quality of R&D outputs because:
Failure leads to feedback to make improvements
Failures initiates the search for the cause of failures
Failures are associated with deliberate learning process (inductively coming up with
innovative ideas)
Failure = rejection of hypotheses, so the next hypothesis can be supported
Failure stimulates distant search outside of knowledge corridors, resulting in more possible
solutions, from which the best one can be chosen
