Presentation is loading. Please wait.

Presentation is loading. Please wait.

Myopia Of Selection: Does Organizational Adaptation Limit The Efficacy Of Population Selection? Hart E. Posen – University of Michigan Daniel Levinthal.

Similar presentations


Presentation on theme: "Myopia Of Selection: Does Organizational Adaptation Limit The Efficacy Of Population Selection? Hart E. Posen – University of Michigan Daniel Levinthal."— Presentation transcript:

1 Myopia Of Selection: Does Organizational Adaptation Limit The Efficacy Of Population Selection? Hart E. Posen – University of Michigan Daniel Levinthal – University of Pennsylvania INFORMS - 2006

2 © 2006 Hart E. Posen 2 Evolutionary Theory: The classic debate.... Significant attention to evolutionary perspectives  Built on the work of Campbell (1965); Nelson & Winter (1982) and Hannan & Freeman (1977) Organizational Adaptation versus Population Selection  Which process best explains average population level performance advancement (i.e. population adaptation)? Organizational adaptation modeled as search resulting in learning  Big literature on the benefits and pitfalls of adaptation as organizational learning Population selection modeled as disproportionate removal of less fit organizations  Literature premised on organizational inertia

3 © 2006 Hart E. Posen 3 What factors that generate superior learning performance? (cf. Argote 1999) Time Performance Significant heterogeneity in learning rates (e.g. Dutton & Thomas 1984)

4 © 2006 Hart E. Posen 4 How Effective is Selection as a Mechanism of Population Change? Time Performance Selection forces Can selection identify of superior learning curves???

5 © 2006 Hart E. Posen 5 Central Premise Selection is systematically error prone  Removing organizations at intermediate points in time that would, had they survived, gone on to do well. Error in selection is endogenous to organizations’ search strategies that guide the learning process Focus on intertemporal reliability  Selection error greatest over learning curves that exhibit lower intertemporal reliability

6 © 2006 Hart E. Posen 6 Examples of Reliability in Learning Curves Short-Wave Reliability Long-Wave Reliability Performance Time (t) T Performance Time (t) T Selection is more error prone when organizations’ search strategies lead to learning curves that exhibit lower intertemporal reliability.

7 © 2006 Hart E. Posen 7 Relation to Prior Literature - 1 Campbell’s arguments about the need for blind variation for effective selection  “if discovery or expansions of knowledge are (to be) achieved, blind variation is requisite”, where blind implies that “specific correct trials are no more likely to occur…than specific incorrect trials” (Campbell, 1982: 86) Efficacy of selection endogenous to nature of organizational variation over time.  Simon dissents: Variation in not blind.

8 © 2006 Hart E. Posen 8 Relation to Prior Literature - 2 -Population Ecology Selection over inert organizations  Organizational inertia (Hannan & Freeman 1977, 1984)  Stable quasi-genetic traits organizational traits Such as routines (Nelson & Winter 1982) Highlights role of reliability – as it relates to survival prospects of individual organizations.

9 © 2006 Hart E. Posen 9 Selection if Fundamentally Myopic Cross-Sectional Challenge (Easy)  Don’t remove superior organizations at a point in time. Longitudinal Challenge (Difficult)  Problem is that selection is fundamentally myopic  Don’t remove organizations that, had they survived, would have gone on to do well. Time Performance Selection forces Under what conditions does myopic selection demonstrate the possibility of prospective intelligence?

10 © 2006 Hart E. Posen 10 Important Questions in Organization Theory Why do inferior organizations (or technologies) sometimes come to dominate the set of survivors?  Outcome is a function of political and social competition as it is technical competition on the merits of alternatives (Anderson, and Tushman, 1990)  Asset stocks act as buffers to stave off failure (Levinthal, 1991; Barnett, Greve, and Park, 1994; Barron, et al., 1994)  Path dependence arises from positive externalities (David, 1985; Arthur, 1989) What is a superior learning strategy?  Long run performance, survival, performance conditional on survival?  e.g. explore versus exploit Exploration is inherently less reliable, but might lead to superior long run performance

11 © 2006 Hart E. Posen 11 The Experiment…. Computational Model  NK (Kauffman 1993, Levinthal 1997, Rivkin 2000) Organizations engage in search and learning  Task environment is nearly decomposable.  Learning curve heterogeneity is driven by different search strategies Multiple experiments  In each experiment, hold search strategies constant, alter the type and strength of selection. Selection occurs throughout learning

12 © 2006 Hart E. Posen 12 Differences in Learning Curves that Result from Heterogeneity in Search Strategies Content Risk (implementation of good alternatives) Process Risk (presence of embedded structures) Strategy Element 1 Mode of Evaluation of Alternatives Strategy Element 2 Implementation of Structure

13 © 2006 Hart E. Posen 13 Adaptation (Learning Curves) in the Absence of Selection Experimental control: All search strategies generate the same average long-run fitness.

14 © 2006 Hart E. Posen 14 Intertemporal Reliability Lowest Short- Wave Reliability Lowest Long- Wave Reliability

15 © 2006 Hart E. Posen 15 Selection Error Rates Lowest long- wave reliability Lowest short- wave reliability Error rates a function of reliability Error rates increase with selection intensity

16 © 2006 Hart E. Posen 16 Fitness Improvement Generated by Selection Search strategies that generate the most reliable learning curves lead to selection that is more effective at improving average population Performance. Most Reliable Search Least Reliable Search Performance Improvement Percent

17 © 2006 Hart E. Posen 17 Error Rate Under Alternative Selection Regimes (History, Discrete) Prior history increases selection error rate (reduces efficiency of Selection) with the exception of on-line search Error exacerbated by discrete selection

18 © 2006 Hart E. Posen 18 Competitive Consequences of Differential Reliability Search strategies that generate the most reliable learning curves lead are more likely to dominate the Population of surviving firms

19 © 2006 Hart E. Posen 19 Conclusion - 1 Selection is systematically error prone. Efficacy of selection is endogenous to the dynamics of organizational adaptation. Question: Under what conditions can myopic selection demonstrate the possibility of prospective intelligence?  Selection is more effective across organizations pursuing strategies that exhibit higher levels of intertemporal reliability. This recognizes the fundamental interrelationship between adaptation and selection.  They are not separate processes

20 © 2006 Hart E. Posen 20 Conclusion - 2 Tragedy of selection  Why do inferior organizations sometimes dominate the set of survivors?  Organization choices that may enhance individual performance may in fact reduce the efficacy of selection and decrease average performance of surviving organizations. Exploration that increases average long run performance, but also decreases reliability. Search strategy that is on average superior also provides the least useful information to intermediate selection. Post selection, firms of this type appear to be, on average, inferior!  Selection is less effective across such strategies.

21 © 2006 Hart E. Posen 21 Thank You


Download ppt "Myopia Of Selection: Does Organizational Adaptation Limit The Efficacy Of Population Selection? Hart E. Posen – University of Michigan Daniel Levinthal."

Similar presentations


Ads by Google