New HRM Practices and Exploitative Innovation: A Shopfloor Level Analysis

Abstract

Extant research documents a positive relationship between the adoption of new human resource management (HRM) practices at the managerial and shopfloor level, and innovation performance, respectively. However, studies focusing on the managerial level distinguish between different types of innovation, while studies at the shopfloor level regard innovation as a homogenous activity. No previous studies have explicitly accounted for innovation heterogeneity in relation to the adoption of new HRM practices at the shopfloor level. Thus, it is still an open question whether and to what extent the findings at the managerial level apply to the shopfloor level. We address this question by focusing on the introduction of exploitative technological innovation and the adoption of new HRM practices at the firm's lowest hierarchical level. We suggest a positive relation between the two via the firm's productive capabilities as employees' ability and learning incrementally improve the firm's existing products and processes. Our argument is successfully tested on a sample of 166 Italian firms.

Keywords:

• New HRM practices
• exploitative innovation
• productive capabilities
• shopfloor work

Acknowledgements

The authors would like to thank the editors and two anonymous reviewers for their constructive comments. The authors are also grateful to Larissa Rabbiosi for insightful suggestions on earlier versions of the paper and to the participants at the 2008 conference on “Organizing for Internal and External Knowledge Creation and Innovation: Looking within or Searching Beyond?” at Copenhagen Business School.

Notes

¹ For a description of NUTS classifications see Eurostat (1995).

² Firm performance data are drawn from the firms' balance sheet submitted at the Chamber of Commerce of Reggio Emilia.

³ These sectors are drawn from an OECD (1994) revision of Pavitt's taxonomy (1984), which intends to aggregate industrial sectors according to market orientations, input characteristics and technological contents for manufacturing firms in order to link sectoral performance with labor markets. No science-based firms were recorded in our sample. However, this does not undermine our analysis since we rely on perceptual measures of innovation.

⁴ The final sample distribution across size and sectors replicates the distribution of the 199 respondent firms across the two dimensions. Specifically, specialized suppliers (39 per cent) and resource-intensive (28 per cent) respondent firms are predominant, followed by labor-intensive (19 per cent) and scale-intensive (14 per cent) ones. In terms of size, respondent firms are mainly small and medium size (43 and 29 per cent, respectively). Thirteen per cent have less than 500 employees, 8 per cent less than 1,000 and the remaining 8 per cent more than 1,000 employees.

⁵ The time series distribution of this variable is characterized by stationarity. We check this property using the Kolmogorov–Smirnov test which reported robust supportive evidence.

⁶ We performed a principal component analysis for the cross-sectional sample that included all explanatory and control variables, and the dependent variable. The analysis retained eight factors with eigenvalues greater than 1.00 with no factor explaining more than 11.29 per cent of the total variance.

⁷ Under the null hypothesis of exogeneity, the coefficient of the residuals of the first stage regression is not statistically different from zero at the second stage.

⁸ This method allows consistent estimates to be generated for non-linear models via Amemiya Generalized Least Squares when addressing estimation bias due to endogeneity and omitted characteristics by instrumenting the independent variables in the model that are thought to be endogenous. STATA ivprobit command (Harkness, 2003) estimates the endogenous variable as a linear function of the instrumental variables and corrects the second step standard errors (Wooldridge, 2002).

⁹ The conditional likelihood-ratio (CLR) and the LM J-tests reject the null hypothesis of exogeneity at p < 0.05.

¹⁰ Since the direct application of Sargan (1958) and Basmann's instrumental variable method (1960) to nonlinear errors-in-variables models fails to yield consistent estimators, Lee (1992) shows that the Newey's minimized distance (or minimum-χ²) for the IV probit estimator provides a test of overidentifying restrictions (1987). Like Sargan and Basmann statistics, the test statistic is distributed as χ² with (L (K) degrees of freedom (where L is the number of instruments, K the number of regressors and L (K the number of overidentifying restrictions) under the null that the instruments are valid.

¹¹ Like for exploitative innovation, we rely on March (1991) to define explorative innovation. We also adopt a firm-centered use of the concepts of exploitation and exploration which allows capturing the fact that activities perceived as exploratory by a firm may be perceived as exploitative by another and vice versa (He and Wong, 2004). Specifically, explorative technological innovation is defined as a new to the firm product and/or process innovation.