The impact of bundles of strategic human resource management practices on the performance of European firms

Abstract

The relationship between HRM practices and perceived firm performance was analysed in 3,281 firms located in European Union countries using data derived from the Cranet data set. A factor analysis of 80 different HRM practices resulted in 15 bundles of HRM practices which were then further categorized as being either ‘calculative’, ‘collaborative’ or ‘intermediary’. While controlling for contingency factors, firm strategy, firm size, market conditions and degree of unionization, as well as controlling for industry and country, the resulting analysis indicates that while five of the six calculative practices and two of the three intermediary practices have a significant impact on performance none of the six collaborative practices has. Significantly it was further noted that the overall effect of HRM on performance was relatively modest.

Keywords:

• calculative HRM bundles
• collaborative HRM bundles
• contingency factors
• human resource management and firm performance

Notes

1. The exploratory factor analysis extracts as many factors as there are variables. Two criteria are used for a parsimonious solution: 1) Kaiser's criterion where an eigenvalue of 1 is the cutting point; 2) an examination of the scree plot and making a cut when there is a significant decline in eigenvalue to the next potential factor and the line in the plot flattens. In our case the next potential factor had an eigenvalue of 1.004, just at the Kaiser's criterion, but there is a significant jump in the eigenvalues from around 1.3 for factor 15 to 1.0 for factor 16. Thus, we decided to accept the 15 factor solution, which is consistent with Stavrou and Brewster (2005). The next step is to examine the pattern of factor loadings. The indicators load strongly on the expected factors with no exceptions as Table 2 shows. The ‘off-diagonal’ factor loadings are omitted from Table 2 for reasons of space. This information is important in evaluating the factor solution. The indicators should ideally show high loadings on the factor, and low loadings on other factors. In fact, there are only two such ‘off-diagonal’ factor loadings that exceed 0.3. One indicator with the highest loading on factor 11 also shows a loading of 0.41 on factor 14. Another indicator of factor 11 loads 0.33 on factor 14.