How Human Resource Management Practices Contribute to Learning for Pharmaceutical Innovation in Mexico: Drawing From Internal and External Sources of Knowledge

ABSTRACT

How and why do human resource management practices contribute to learning for innovation? Based on evidence from the Mexican pharmaceutical industry, we found that the influence of human resource management strategies on learning processes is more perceptible as learning becomes more exploratory in nature. The provision of training and, to a lesser extent, worker empowerment and the regulation of hiring practices underpin learning for pharmaceutical innovation. However, the influence of firm's characteristics on the adoption of strategies to learn from internal and external sources of knowledge tends to outweigh that of the human resources management practices of the firm.

RESUMEN

¿Cómo y por que las prácticas de gestión de los recursos humanos contribuyen en el aprendizaje de la innovación? Con base en evidencias obtenidas de la industria farmacéutica en México, observamos que la influencia de las estrategias de gestión de los recursos humanos sobre los procesos de aprendizaje se percibe mucho más a medida que la naturaleza del aprendizaje se torna más exploratoria. El entrenamiento brindado y, en menor medida, el empoderamiento del trabajador y la regulación de las prácticas de contratación respaldan el aprendizaje para una innovación farmacéutica. No obstante, la influencia de las características de la empresa en la adopción de estrategias para aprender de las fuentes de conocimiento internas o externas, tiende a superar las inherentes a las prácticas de gestión de los recursos humanos de la compañía.

RESUMO

Como e por que as práticas de gestão de recursos humanos contribuem para o aprendizado da inovação? Com base em evidências da indústria farmacêutica no México, descobrimos que a influência das estratégias de gestão de recursos humanos nos processos de aprendizagem é mais percebida à medida que a natureza do aprendizado torna-se mais exploratória. O fornecimento de treinamento e, em menor grau, o empoderamento do trabalhador e a regulamentação das práticas de contratação servem de base para o aprendizado para a inovação farmacêutica. Porém, a influência das características da empresa na adoção de estratégias para o aprendizado a partir de fontes de conhecimento internas e externas tende a sobrepujar a influência das práticas de gestão de recursos humanos da empresa.

KEYWORDS:

• capability building
• human resource management
• learning
• Mexico
• pharmaceutical innovation

ACKNOWLEDGMENTS

Earlier versions of this article benefited from comments by Ludovico Alcorta, Pierre Mohnen, two anonymous reviewers, and participants from the 10th Globelics International Conference, Hangzhou, China, 2012, and an anonymous reviewer for this Special Issue of the Latin American Business Review. I am greatly indebted to the following at Instituto Nacional de Estadística, Geografía e Informática (INEGI), Mexico, for granting access to the data used in this article: Gerardo Leyva and Abigail Durán. Special thanks to Adriana Ramírez, Gabriel Romero, and Cándido Aguilar at INEGI for help in the processing of the data.

This piece of research was made possible thanks to funding from UNU-MERIT and Mexico's Council for Science and Technology (CONACYT 128043).

Notes

Source: Author with information from ENESTYC (2010).

Note. Categories are exclusive; sample without missing values, n = 112;.

***significant at 1% level of confidence; percentage share of each cell relative to total firms in sample within parentheses.Source: Author with information from ENESTYC (2010).

*Refers to the minimum and maximum average remuneration in MX $1000; sample without missing values = 112.

Source: Author with information from ENESTYC (2010).

Note. a. rd_inhouse, c. rd_improve_process, d. rd_design_meq, e. rd_drug_improvement, f. rd_drug_design, b. external_mkt. + interaction term between internal and external training. The model for the pair b|d in the third column includes only the interaction term for training because the variable training_internal tended to predict perfectly the probability that a firm performs R&D for design of new drugs; STATA drops the problematic variables out from the equation; Sample without missing values = 112; ***, **, *significant at the 1%, 5%, and 10% confidence level; robust standard errors in parentheses; Degrees of freedom within squared brackets; Marginal effects correspond to changes in conditional probabilities that a firm adopts an external learning strategy provided that it already carries out in-house R&D; these are probabilities for combinations of (external = 1|internal = 1).

Source.: Author based on information from ENESTYC (2010).

Note. g. rd_exploit, h. rd_explore, i. extrel_rd, b. external_mkt; Sample without missing values = 112; ***, **,*significant at the 1%, 5%, and 10% confidence level; robust standard errors in parentheses; Degrees of freedom within squared brackets; Marginal effects correspond to changes in conditional probabilities that a firm adopts an external learning strategy provided that it already carries out in-house R&D.

Source: Author with information from ENESTYC (2010).

Note. Figures in bold are significant at 0.05% of confidence or less.

Source: Author with information from ENESTYC (2010).

The perceived innovation crisis in pharmaceutical innovation is not without its critics; see Light and Lexchin (2012) and the ensuing debate in the British Medical Journal.

Rothaermel and Deeds (2004), Gilsing (2006), and Kettler and Modi (2001) used similar interpretations in studies about biotechnology and pharmaceutical firms.

The results from models with a single variable on the provision of training were not significantly different compared to those presented here. However, the distinction between types of training helped to avoid specification troubles preventing the models to converge.

Based on Cassiman and Veugelers (2006), each pair of internal and external learning strategies was regressed on the control variables only. Then we ran full specification models and compared results—this procedure is an indirect way to test for complementarity between learning strategies. Building on the basic structure of the models, an additional bivariate specification included only those explanatory variables that showed statistical significance at the five per cent level or lower in the basic model. For reasons of space and comprehensiveness of the analysis, we omit presentation of results from those equations.