HRM-performance research: under-theorized and lacking explanatory power

Abstract

The search for a measurable link between HR practices and organizational performance is currently preoccupying HR professionals, consultants, government and academics. Empirical research on this human resource management-performance (HRM-P) link is, however, marred by a serious problem: it is under-theorized. While some (but by no means all) researchers on the HRM-P link are aware of the problem, none are prepared to face up to the scale of the implications. Without theory, research on the HRM-P link lacks explanatory power. The only ‘solution’ on offer (the assertion that theory will develop via more and/or better empirical work) has been less than successful: empirical research has multiplied with little or no theoretical development. Nor can it. The under-theorization and lack of explanatory power is rooted in the ‘scientific’ perspective that underpins empirical research. The paper draws upon critical realist philosophy to reveal exactly why the ‘scientific’ approach encourages under-theorization and lack of explanatory power and, furthermore, why the ‘solution’ on offer cannot solve the problem. The conclusion notes why the HR community should not avoid philosophical issues.

Keywords:

• Critical realism
• ontology
• epistemology
• causality
• theory
• explanation

Notes

 1 See, for example, Accenture (2004), Bukowitz et al. (2004), Donkin (2002, 2003, 2004), Kingsmill (2003), Overall (2002), PriceWaterhouseCoopers (2003), Saratoga (2005), Thomas et al. (2003), Witzel (2004), Wright (2001), Woodrow Wyatt (2001/2: 5 and 1).

 2 Rather than provide a long list of scores of these oft-cited references, we refer the reader to two recent surveys Wall and Wood (2005: 454) and Boselie et al. (2005: 81–2). It should be noted that Wall and Wood conclude that ‘existing evidence for a relationship between HRM and performance should be treated with caution (2005: 453). In another survey, Godard (2004: 355) writes: ‘Overall, these concerns suggest that we should treat broad-brush claims about the performance effects of [HRM practices], and about research findings claiming to observe them, with a healthy degree of scepticism.’ The empirical evidence of an HRM-P link is, at best, inconclusive.

 3 See also Becker and Gerhart (1996), Guest (1997, 2001), Guest et al. (2003), Guest et al. (2004), Laursen (2002), Toulson and Dewe (2004), Wright et al. (2003). It is possible to find articles that start by lamenting the lack of empirical and theoretical work on the HRM-P link, and end by quietly abandoning the theoretical dimension, leaving the empirical dimension as an (inadequate) substitute (e.g. Benkhoff, 1996). This is a point noted by Haynes and Fryer (2000: 242). In the introduction to a series of case studies, Becker and Huselid claim to provide ‘an outline of the theoretical rationale and empirical literature linking HRM systems with corporate performance (1999: 288). We see little that can be described as ‘theoretical rationale’.

 4 While some recognize technical problems with metrics, measurement and data, (cf. Becker and Gerhart, 1996; Gerhart, 1999; Gerhart et al., 2000; Wright and Sherman, 1999), we do not address these problems because our critique is aimed at the philosophical argument which overarches these technicalities.

 5 See also Boudreau and Ramstad (1999), Budhwar (2000), Guest (1997, 2001), McMahan et al. (1999).

 6 Two exceptions are Wright and MacMahan (1992) who do address philosophy and the HRM-P link; and Kane (2001) who considers philosophy in HRM more generally. The Human Resource Management Review (Steel, 2003) has a symposium devoted to ‘methodological issues in absenteeism research’ (emphasis added). While this is a slightly different subject matter, there are lessons for us. The symposium does not address methodological issues beyond problems of quantification and research design. All articles in the symposium operate, unquestioningly, with a ‘scientific’ approach. The tenor of the symposium might be grasped from the opening comment. When the editor ‘invited me to serve as guest editor on a special issue devoted to methodological issues ... I immediately decided that the issue's panelists should be drawn from among the ranks of the discipline's most active and prolific researchers. No keener insight into methodological issues is attainable than that won on the empirical battlefield where our most careful and painstaking efforts are so often held hostage to the whims and vagaries of the methodological equivalent of the Greek Fates’ (Steele, 2003: 153). A missed opportunity?

 7 For an introduction to critical realism see Archer et al. (1998), Carter and New (2004), Danermark et al. (1997), Lawson (1997, 2003), Reed (2001), and Sayer (1994, 2000). For a discussion in management studies more generally, see Ackroyd and Fleetwood (2000), Fleetwood and Ackroyd (2004) and Johnson and Duberly (2000: chapters 2 and 3).

 8 We should note, first, that we are fully aware of, let us say, ‘political’ critiques of HRM in general, and of managerial accounts of high performance work systems, deriving, for example, from Labour Process Theory, Critical Management Studies and Interpretive or Phenomenological approaches (e.g. Bowman and Ambrosini, 2000; Ramsay et al., 2000; Sandberg, 2000). We are also aware of critiques coming from, let us say, postmodern, poststructural, social constructionist and/or linguistic or discursive perspectives (e.g., Evans, 1999; Harley and Hardy, 2004; Keenoy, 1997; Legge, 2005; Watson, 2004). While we have sympathy with much of this critical work, we are engaged in a different form of critique, namely, a critique aimed specifically at philosophy. We do, however, have significant differences with the ontological commitments of many (but not all) of these writers. We worry that a strong social constructionist ontology can result in philosophy of science collapsing into sociology of science, making it impossible to even discuss issues of what a plausible theory or explanation might look like (cf. Fleetwood, 2004, 2005; Potter and Lopez, 2001). Second, we exclude from our critique, those writers who do attempt to explain, without being preoccupied with metrics and measures , the nature of the causal mechanisms and processes that may govern any alleged empirical HRM-P link (Bowen and Ostroff, 2004; Dooreward and Melhuizen, 2000; Edwards and Wright, 2001; Elias and Scarbrough, 2004; Folger and Turillo, 1999; Hendry et al., 2000; Knox and Walsh, 2005; Truss, 2001). This approach conforms, at least at the meta-theoretic level, to what critical realists refer to as a causal-explanatory method (Fleetwood, 2004).

 9 References are often made about testing the prediction, testing the hypothesis, testing the theory, testing the model, testing the model's predictions, finding the predictors of their dependent variable and so on. The terminology varies and is highly ambiguous.

10 In a recent survey of 467 articles on HRM, Hoobler and Brown Johnston (2005) not only found just one article on philosophy, they also found that: ‘statistical regression was by far the method of choice, represented in a full 35 percent of the articles studied. Various analysis of variance and meta-analysis accounted for 9 percent and 5 percent respectively, while correlation and structural equation modeling or confirmatory factor analysis respectively amounted to 6 percent and 3 percent’ (2005: 668). The preoccupation with quantitative analysis is also demonstrated by Boselie et al. (2005: 70).

11 A deterministically closed system can be expressed probabilistically and can, thereby, be transposed to a stochastically closed system. It is important to note, however, that closed systems do not cease to be closed systems simply because we specify them stochastically. Under stochastic closure y and x ₁,x ₂…x _n are still constantly conjoined, albeit under some well-behaved probabilistic function. In effect, the claim ‘whenever event x then event y’ is transposed into the claim ‘whenever events x ₁, x ₂,…x _n on average, then event y on average’, or ‘whenever the average value of events measured by variables x ₁, x ₂,…x _n are what they are, then the average value of event y measured by variable y is what it is’.

12 Similarly, summarizing the findings of three leading studies, Gerhart (1999: 32) observes that ‘a one standard deviation increase in various HRM measures is associated with profits (return on assets) that are higher by 23, 16 and 23 percent respectively’.

13 It is important to tackle the red herring here. It is true that the list of what could, in principle, be included in a robust explanation, could easily expand until it included, literally, everything, and go all the way back to the big bang. In practice, however, social scientists usually avoid a potential infinite regress by making use of abstraction (cf. Sayer, 1998). That is, they make judgements about which factors need to be included and which can safely be excluded. This is of course fallible, and sometimes investigators get it wrong – but it is, in principle, no different than deciding upon which variables to include and which to exclude (cf. Runde, 1998).

14 Notice that critical realism advocates using hermeneutic and/or other subjective or interactive research techniques. Furthermore, critical realism can also live with deconstructive techniques aimed at uncovering the exercise of power or revealing how power is used to create interpretations of the world that are presented as true, even if they are false.

15 There are a few researchers who claim to do something similar, but their talk of ‘examining the mechanisms’ quickly evaporates into measuring the mechanisms (Meyer and Smith, 2000: 319). The same goes for Becker and Huselid (1999: 288) as their talk of providing us with ‘rich detail in how leading firms use their HRM system’ and ‘an outline of theoretical rationale and empirical literature linking HRM systems with corporate financial performance’ turns out to be yet more measurement.

16 Den Hartog and Verburg's (2004) attempt to deal with the hermeneutic issue of awareness of organizational culture soon collapses into attempts to measure culture.

17 See also the forum in Administrative Science Quarterly introduced by Sutton and Straw (1995), and the symposium in Academy of Management Review introduced by Ven de Ven (1989).

18 Apart from Wright and McMahan (1992) we could not find any examples from the HRM-P literature.

19 Many researchers get embroiled in various contradictions that spring from the attempt to remain within, and to go beyond, the ‘scientific’ approach. Lahteenmaki et al. (1998) make use of a range of statistical techniques in their search for positive relations between strategic HRM and company performance. That they identify hardly any such relations is besides the point here as they use the same kind of empirical approach as those that do claim to find positive relations, which implies they accept their usefulness. Yet they end up raising doubts about these same techniques and call for intensive case studies to reveal the complex nature of the relationships. Despite their obvious desire to go beyond the current state of affairs, Peccei and Rosenthal (2001: 838) cannot break with the ‘scientific’ perspective and end up ‘first modelling and then testing the core assumptions linking HR practices and management behaviours to the customer orientated behaviour of front-line workers’. While much depends upon the methods adopted, many who opt for a mixed-methodology strategy (e.g. Budhwar, 2000), actually end up with what amounts to the ‘scientific’ approach with some (often useful) insights bolted on.