Consultation and communication in family businesses in Great Britain

Abstract

Nationally representative data on family businesses are available in the 1998 Workplace Employee Relations Survey, alongside comparable information for other types of firms. We use these data to compare differences in the use of different consultation and communication procedures. We cover such practices as the use of direct communication schemes (e.g. briefings; the provision of information on financial performance to the workforce) as opposed to indirect methods such as the use of joint consultative committees. There is an a priori expectation in the literature that family-owned businesses are either more likely to use direct forms of communication (vis-à-vis indirect forms) or that they will not be involved in direct communication or consultation with their employees, and we test this using multivariate techniques. Finally, we consider whether the type of consultation/communication structure matters in terms of establishment performance, and what differences exist with respect to family-owned businesses. In particular, this paper tests if those firms that consult directly with staff, as opposed to those that consult through joint consultative committees or trade unions, have higher productivity and/or other measures of performance. Concurrently we test whether there are separate ‘family business’ effects or whether it is generally establishment size that ‘matters’, by estimating a model for family-owned and non-family-owned establishments. In general, our results show that not only do family-owned establishments have lower levels of communication and consultation, but, when the latter is present, this does not generally translate into greater economic benefits (as is the case in non-family-owned firms).

Keywords:

• Employee involvement
• family-owned firms
• performance

Notes

Renee S. Reid (address for correspondence), Director, Caledonian Family Business Centre, 4th Floor, Hamish Wood Building, Caledonian Business School, Glasgow Caledonian University, Cowcaddens Road, Glasgow. Scotland (tel: [phone]; fax: [phone]; e-mail: [email protected]). Professor Richard I.D. Harris, Professor of Economics, University of Newcastle Business School. Dr Rodney McAdam, Reader, School of Business Organisation and Management, University of Ulster at Jordanstown.

1 An early comprehensive review of family business literature and publications, Desman and Brush (1991) , reported that only 4 per cent of the 202 citations reviewed dealt with the development of human resources through education and training. In the period since this review, little additional research into HRM in family businesses has been undertaken.

2 There is a debate about how best to define a family-owned vis-à-vis non-family-owned firm (see, for example, Daily and Dollinger, 1993 ). In this study we are constrained by the question asked in the 1998 Workplace Employment Relations Survey (if the establishment was in the private sector but was not a PLC, it was asked: ‘does a single individual or family have controlling interest over this company, where controlling interest means at least 50 per cent ownership?’).

3 See also Wood et al. (2002) ; they provide empirical evidence of the link between what are termed high-involvement HR systems and organization performance, as do Arthur (1994) , MacDuffie (1995) , Huselid (1995) , Becker and Huselid (1998) , Appelbaum et al., (2000) .

4 Freeman et al. (2000) include the following as examples of EI practices: the extent of self-managed work teams, worker involvement in the design of EI programmes, the extent of TQM, committees on productivity, worker involvement in work processes, formal suggestion or complaint systems, formal information-sharing with employees and surveys of workers regarding their satisfaction.

5 We recognize that there is a debate concerning the extent to which employee involvement (and its constituent elements) leads to increased employee loyalty, responsibility and effort, and thus increased efficiency, or whether these measures actually lead to ‘work intensification’ operated through a different form of employer control (cf. Marchington and Grugulis, 2000 ). A fundamental issue is whether EI leads to greater employee empowerment (Wilkinson, 1998 ) or whether it is a mechanism for the appropriation of employee knowledge and while giving workers greater specific control over their day-to-day tasks nevertheless allows management greater general control over the work system as a whole. This is an important debate, but if the outcome is still greater productivity of the workforce (through empowerment or greater control), then it might be argued that EI initiatives can still achieve their goals.

6 Although, see note 5 above which points out that there is a debate as to whether greater EI leads to more empowerment of workers or rather greater work intensification. Limiting EI to greater consultation and communication, it does seem likely that this offers more scope for attempting greater rent-seeking.

7 More generally, researchers in family business believe that family involvement makes a family business distinct from a non-family-owned business, but ‘unfortunately, our understanding of the nature of this distinction and its impact on firm performance is incomplete’ (Chua et al., 2003 : 331). One of the aims of this study is to test whether being family-owned is empirically important and to try to identify the channels through which such firms operate differently (and the consequences that arise).

8 This contingency-based approach will be tested later by introducing our variables for consultation and communication in conjunction with a measure of internal fit, to see if the former impact on performance only when introduced as composite (i.e. interactive) variables involving internal fit.

9 Note, WERS98 covers only establishments employing ten or more employees; micro-firms are therefore omitted and this is likely to explain the fact that surveys that include smaller firms tend to find that family-owned establishments are on average older (see Barclays Bank, 2002 ).

10 WERS98 collected information on the number of full-time males (and females) that earned in six different earnings bands (from <£9,000 to £29,000 or more). This information was used to calculate a Gini coefficient that measures the relative inequality of the wage distribution for each establishment. Note, there are differences in the relative value of the Gini coefficient across industries; e.g. in distribution, hotels and restaurants (where family-owned plants are concentrated), the coefficient for family-owned establishments is 0.35 compared to 0.40 in non-family-owned plants; in agriculture, the relative values are 0.22 and 0.33, respectively. However, in almost every other sector the Gini coefficient is larger in the family-owned sub-group.

11 Although not shown in Table 1 , family-owned plants also had more male full-time workers earning less than £9,000 (21 per cent as opposed to 13 per cent in non-family-owned plants) and fewer earning £29,000 or more (7 per cent compared to 14 per cent). T-tests that these means are statistically different are highly significant at better than the 1 per cent level.

12 Note: we recognize that there can be quite important differences between employee information-sharing, communication and consultation practices as these relate to empowerment of employees. However, as argued by Wilkinson (1998) , management increases downward communication typically via newsletters, the management chain or team briefing, which should result in greater employee commitment. Upward problem solving (through say the use of quality circles or suggestion schemes) should also have similar impacts. What is important to distinguish is whether either (1) such practices are absent or (2) communication and consultation is direct or mediated by employee representation (i.e. JCCs). Hence, our grouping of practices in Table 2 into those presented in Table 3 .

13 Note: when workplaces belonging to enterprises employing more than 500 employees are omitted (i.e. just using the SMEs in the WERS database), we still get a similar set of significant results vis-à-vis those reported in Table 2 . Note that Table 2 (last column) controls for size differences in any event.

14 Note: all marginal effects are calculated at the mean of the variables concerned.

15 As expected, nearly all family-owned plants that had no direct communication/consultation also had no trade union members, although interestingly 5.3 per cent of such plants did.

16 There are potential problems associated with bias when using such ‘perception’-based data. However, Wood et al. (2002) state that if there is bias then as long as it is uniform this will not invalidate the measures used. In addition, they report that correlations among the various performance measures are low (respondents were not overly optimistic in any systematic sense); they also found no bias in performance results between workplaces that were involved in benchmarking and those that were not.

17 Since the data used are cross-sectional, the results produced cannot be used to substantiate any causal relationships, only associations between the variables used (although in most cases we believe that the dependent variable – financial performance – is likely to be causally determined by the regressors we include in the model).

18 An alternative approach would be to estimate separate equations. However, this reduces the sample size through sub-dividing the data (especially when – as will be seen – we wish to allow other variables such as external and internal fit to act as intermediaries in terms of how other variables impact on performance) while pooling potentially separate equations also allows us to test the significance of any differences directly (through whether composite variables are statistically different from zero or not).

19 All these extra variables were also entered again multiplied by the family-owned dummy to ensure we captured all potential channels for contingency effects.

20 See Greene (2000 , especially pp. 877–8), who states: ‘it is quite unclear how the coefficients in the ordered probit model should be interpreted’.

21 The ‘external fit’ variable itself was entered in a stepwise approach and, apart from its featuring in one statistically significant composite term (discussed below), external fit was found not to be a significant determinant of financial performance.

22 Also compare the parameters reported in Table A2 for the direct communication variables, which have almost equal but opposite values.

23 Some 2.6 per cent of family-owned workplaces had no communication with their workforce and targeted the quality of the product or service.