High involvement work systems and economic performance: a review of recent research
Many commentators suggest that "high involvement work systems" may markedly improve organisational performance. Drawing upon UK research to be published later in 2000, this feature reviews and critically examines these claims.
Although definitions of "high involvement work systems" (HIWS) vary widely, most writers on the issue emphasise the active participation of employees in the work process through self-managed teams and problem-solving groups, together with arrangements for information-sharing between management and labour, and employee training and skill development. A growing body of studies suggest that these systems may raise organisational performance.
One recent article, for example, maintains that "enormous economic returns [can] be obtained through the implementation of what are variously called high involvement, high performance or high commitment management practices" (see "Putting people first for organisational success", J Pfeffer and JF Veiga, Academy of Management Executive (1999)). Some researchers have gone so far as to estimate precise financial returns from these forms of work organisation (for example, "The impact of human resource management practices on turnover, productivity and corporate financial performance", M Huselid, Academy of Management Journal (1995)). Other studies point to an association between the use of high involvement practices and outcomes including reduced scrap and waste, improved productivity and product quality, and greater process continuity or reliability.
These debates are not confined to the academic sphere but are increasingly influential in public policy and management practice. Bodies as diverse as the European Commission, the Organisation for Economic Cooperation and Development and the American trade union federation AFL-CIO have all, in recent years, urged the take-up of HIWS as a means of promoting productive efficiency with socially desirable goals, such as employee participation and consultation. Business magazines aimed at practising managers, such as the Academy of Management Executive and the Harvard Business Review, have also carried articles advocating the adoption of these practices.
The claims made for HIWS have recently been criticised on several fronts (for example, "Best practice and best fit: chimera or cul-de-sac?", J Purcell, Human Resource Management Journal (1999)). Drawing on yet to be published research carried out at the University of Warwick's Industrial Relations Research Unit ("High involvement work systems and performance outcomes: the strength of variable, contingent and context-bound relationships", P Edwards and M Wright M, International Journal of Human Resource Management (forthcoming)), below we highlight three issues in particular.
First, HIWS are often poorly specified. High involvement systems are often contrasted with allegedly rigid "Fordist" systems of tight supervision and fragmented job tasks. There is less agreement as to their positive features. One article examined five recent studies in the USA and no found that no single measure of HIWS was common across all the studies and only two (presence of self-directed teams and of problem-solving groups) were present in four out of five. Even with these core practices, there is considerable scope for definitional dispute. Ascertaining the degree to which a workgroup is "self-managed" is likely to be problematic, since the boundary of control between the group and management may be continually renegotiated and therefore constantly shifting. Consequently, even similarly constituted workgroups may vary greatly in their autonomy from first-line management, and in the range of issues (eg work scheduling, the required level of labour effort, labour selection or the purchase of new equipment) over which they exercise a measure of discretion.
An equally neglected area concerns organisational performance. Few of the available studies discuss what is meant by this term; most articles appear uncritically to adopt one or more measures with little or no justification as to their choice. An immediate concern is that studies may be measuring quite different things under the positive-sounding label of "performance".
A second issue is that HIWS generate a large number of outcomes, with the risk of selectivity bias in concentrating upon a small number of potentially unrepresentative gauges. For example, direct measures of labour behaviour, such as the level of employee absenteeism, or the incidence of industrial disputes, may be critically important standards of performance, as may production outcomes, such as the level of output per worker, or the rate of defective goods produced. These must be set against technical efficiency issues, such as the efficiency of use of raw materials or the reliability of equipment and machinery, and also against financial outcomes, such as return on average capital employed, or baseline profitability. Many other indicators might also be cited: the Canadian aluminium multinational, Alcan, records over 100 measures in each of its smelters worldwide at monthly intervals.
Moreover, such indicators are multi-faceted, in that an upward shift in one may precipitate a downward movement in others. For example, a rise in output per person may occur at the expense of an increase in scrap or a deterioration in quality. Machinery failure, through greater wear and tear, may also worsen. As an illustration, the relative performance of two plants of the engineering multinational ABB altered greatly according to various ABB measures. In short, HIWS may give rise to a multiplicity of outcomes, which may shift in mutually opposed directions and which must be interpreted in context, rather than selectively or in isolation.
The third issue refers to the nature and direction of any association between HIWS and performance. It is commonly assumed that high involvement systems exert a direct and positive effect upon outcomes. This, it is held, occurs through increased labour effort, better training and the exercise of greater diligence by employees in the execution of their tasks. However, few studies are able to demonstrate empirically such a link. Many lack data on such unobtrusive matters as the level of labour effort. Most studies are cross-sectional, and present a set of associations at a single point in time. And, of course, establishing a correlation between two or more variables does not prove causality.
How far quantitative analyses are able to capture subtle interactions between structures, actors and contexts is also open to question. As such, HIWS may not exert the effects that are claimed for them. HIWS may simply remove the rigidities associated with previous production regimes. Equally, where restructuring occurs at a time of crisis and job shedding, as seems common, employment insecurity may be driving changes in employee behaviour, rather than manifest features of the work system. The effects of competitive pressures and production system redesign may also be so closely entwined as to be impossible to disentangle.
One final point is that the direction of causation may be the reverse of what is commonly assumed. Firms with greater than average profits may be better placed to make the necessary investments in training and development, and to withstand any temporary downturn in revenues or other disruption incurred during reorganisation.
The relationship between HIWS and performance is less direct than is often portrayed, and is mediated by a complex set of variables. Relevant intervening factors include labour attitudes and behaviour, and available production technology and its configuration. The external environment, especially prevailing product market and labour market conditions, frames these. Viewed as a network of links between a set of inter-related variables, which are themselves embedded within a specific context, the relationship between HIWS and performance outcomes is less likely to be one of strict determinism than of contingency and reciprocal conditioning.
Many writers on HIWS and performance have argued that effects are universal (ie applicable to all cases), direct (clear and tight causal links are assumed), one-way and unambiguous in respect of possible outcomes. A considered appraisal suggests that HIWS are often poorly specified, performance is multi-dimensional, and any causal links are weaker and more context-dependent than is generally held. (Martyn Wright, IRRU)