Title:
Helpfulness and Productivity: Implications of a New Taxonomy for Star Scientists
Helpfulness and Productivity: Implications of a New Taxonomy for Star Scientists
No Thumbnail Available
Author(s)
Oettl, Alexander
Advisor(s)
Editor(s)
Collections
Supplementary to
Permanent Link
Abstract
The need to hire the best and the brightest - "the war for talent" - has long been one of the most pressing strategic concerns facing managers (Kapur and McHale, 2005; Guthridge, Komm, and Lawson, 2008). This concern is largely driven by the observation that high performers, or stars, account for the generation of a disproportionately large level of output. The vice-president of engineering of Google, Alan Eustace, noted to the Wall Street Journal in 2005 that "one top-notch engineer is worth 300 times or more than the average", and that he "would rather lose an entire incoming class of engineering graduates than one exceptional technologist" (Tam and Delaney, 2005). Why is this? How do stars so greatly influence the performance of organizations? The existing performance taxonomy for scientists focuses exclusively on individual output, classifying a scientist as either a Star or a Non-Star. The seminal work by Zucker, Darby, and Brewer (1998), for example, defines stars as the top 0.75% of contributors to the genetic sequence database GenBank, a group that accounts for almost 17% of contributions. Recent work by Groysberg, Lee, and Nanda (2008) examines the skill portability of the top 3% of security analysts when they move firms, using a ranking of the perceived effectiveness of security analysts and Azoulay, Graff, Zivin, and Wang (2008) look at the impact of eminent scientists using a variety of measures; such as research funding, citations, and patenting. In all of these articles, the definition of a star is based solely on productivity, in other words, we define stars by what they physically produce. This uni-dimensional classification of star scientists is surprising as innovation is most often characterized as a communal process. Communal interactions matter for two reasons. First, innovation is more often a result of the recombination of existing knowledge and ideas, rather than the discovery of something fundamentally novel (Gilfillan, 1935; Nelson and Winter, 1982). As knowledge frontiers continue to expand, combinations of increasingly specialized levels of human capital are required to reach the forefront of knowledge (Wuchty, Jones, and Uzzi, 2007; Jones, 2008). It is this recombination of specialized ideas, either through formal collaborations (coauthorships, joint ventures, etc.) or informal means (discussions and comments from helpful individuals), that leads to innovation. Second, the exchange of knowledge is to a large extent governed through social channels. Individuals possess only finite levels of knowledge and knowledge search is costly; social forces can reduce barriers to knowledge flow through geographic proximity (Jaffe, Trajtenberg, and Henderson 1993), labor mobility (Almeida and Kogut, 1999; Oettl and Agrawal, 2008), social networks (Singh 2005), and membership in ethnic communities (Agrawal, Kapur and McHale 2008). While innovation is a communal process, the inability to perfectly contract between parties on knowledge exchange leads to failures in the market for knowledge and a decrease in knowledge transfer (Arrow 1962). As such, conditions that facilitate knowledge sharing or spillovers in the absence of formal contractual environs are of great value to firms. Ultimately, if our concern is to understand the mechanisms by which an individual maximizes his performance, simply understanding the productivity inputs of an individual would suffice. However, the strategy and economics literatures focus on performance measures at the organization and regional levels, and as such, mechanisms in which individuals influence the productivity of others become important as these mechanisms directly influence the performance of organizations and regions. Hence, mechanisms by which individuals generate spillovers are of paramount concern to scholars of strategy and economics. The importance of social factors on innovation illuminates the deficiency of our current productivity-focused conceptualization of star scientists (Stars versus Non-Stars). To expand our current conceptualization of star scientists, I develop a new taxonomy of star scientists by incorporating a social dimension: helpfulness to others. This new taxonomy allows an individual to not only vary along a productivity dimension but also along a helpfulness dimension. The objective of this paper is threefold. First, I expand upon the current dichotomous conceptualization of stars by developing a taxonomy that not only incorporates a star's individual productivity but also his helpfulness. In doing so, I move beyond the current uni-dimensional classification and redefine what it means to be a star. Second, I propose a measure to classify individuals into this new taxonomy. And third, I use this taxonomy to assess the extent to which different star types influence the productivity of others. Following prior studies (Allison and Long, 1990; Azoulay, Graff, Zivin, and Wang, 2008) I measure individual productivity using Impact Factor-weighted publication counts. Helpfulness, on the other hand, is measured by academic journal acknowledgements as acknowledgements are generally made to those who have helped in the development of the work. Using these measures of productivity and helpfulness, I classify a sample of 415 immunologists and examine their influence on the productivity of their coauthors. Coauthorship is used to pinpoint the timing of the formation of an interpersonal tie between the immunologist and a potential recipient of spillovers. It is this collocation in social space that allows the coauthor the potential to benefit from any spillovers the star may provide. By placing a star in both productivity and helpfulness space, while keeping the classifications discrete, I am able to classify an individual as one of four types: an All-Star, a Lone Wolf, a Maven, or a Non-Star. I define an All-Star as an individual with both high productivity and high helpfulness. A Maven is an individual with average productivity but high helpfulness. A Lone Wolf is someone who has high productivity but average helpfulness, and a Non-Star has both average productivity and average helpfulness. Restrictively, the current dichotomous conceptualization of stars groups both All-Stars and Lone Wolves together, and overlooks Mavens. By expanding on the current classification, I am able to examine the influences of individuals who vary both in their productivity and their helpfulness. Examining the changes in productivity from coauthoring with various star types would be an appropriate empirical exercise if coauthoring relationships were chosen at random, but clearly they are not. The problem with endogenous coauthor selection is that the coauthors selected by an immunologist may be chosen due to their productivity, thus producing spurious correlations between an individual's productivity and their coauthorship network. For this paper, I examine the decrease in productivity of coauthors when an immunologist dies. Across a number of specifications, the productivity of the coauthors of All-Stars that die decreases on average by 35% relative to the decrease in productivity when a Non-Star dies. More interestingly, coauthors of Mavens that have died experience a 30% decrease in productivity, while the coauthors of Lone Wolves experience decreases in productivity of only 19% on average. By expanding the current conceptualization of star scientists and focusing on both the productivity and helpfulness dimensions of scientists, I find that spillovers are most likely to be generated from individuals with high helpfulness. As a result, the literature has largely overemphasized the importance of Lone Wolves, yet overlooked and consequently underemphasized Mavens.
Sponsor
Date Issued
2009-10-03
Extent
Resource Type
Text
Resource Subtype
Proceedings