A decision model for outsourcing training functions: distinguishing between generic and firm-job-specific training content

Abstract

Employee training plays a crucial role in the success of most organizations. Due to its developmental aspect, training is closely linked to core competencies and strategic focus. However, it is also one of the most widely outsourced HR functions in most Western economies.

This article attempts to illustrate and propose a decision model for the factors that shape the expected benefits and subsequently the extent of outsourcing training functions. A distinction is made among generic training (for the development of competencies) and job- or company- specific training (for example, induction training, job specialization, etc.).

Two decision models are extracted with structural equation modelling. Asset specificity, market availability, in-house development of training and firm size are discussed. The factors shaping the decision to outsource, as well as the perceived benefits from outsourcing employee training, are different for each of the two types of training (generic and specific). The reasons underlying those differences are discussed. For both types of training service it is proposed that the expected quality benefits, not cost ones, induce companies to outsource training.

Thus, this study attempts to offer a useful insight into the factors shaping the extent and the expected benefits from outsourcing training services. The outcomes can further assist HRM professionals (managers and providers of HRM services), as well as academics to gain a better understanding of the nature of HRM outsourcing decisions in general, and a ‘basic’ HRM outsourcing practice – training – in particular.

Keywords:

• core competencies
• decision making
• Greece
• training and development
• transaction costs

Notes

1. Normative Fit Index: is one of the most widely used indices of fit in Structural Equation Modelling and it shows how well the data fit to the hypothesized model (Byrne 2001).

2. Comparative Fit Index: this index of fit is taken by NFI, but it takes sample size into account, so that fit is not underestimated in small sizes.

Values for NFI and CFI range from zero to 1 and a value > 0.90 is generally considered representative of a well-fitting model (Byrne 2001).