HRM and Small-Firm Employee Motivation: Before and After the Great Recession

Abstract

A long-running debate in the small-firms’ literature questions the value of formal human resource management (HRM) practices, which have been linked to high performance in larger firms. The authors contribute to this literature by exploiting linked employer–employee surveys for 2004 and 2011. Using employees’ intrinsic job satisfaction and organizational commitment as motivational outcomes, the authors find the returns to small-firm investments in HRM are U-shaped. Small firms benefit from intrinsically motivating work situations in the absence of HRM practices and find this advantage disturbed when formal HRM practices are initially introduced. Firms can restore positive motivation when they invest intensively in HRM practices in a way that characterizes high performance work systems (HWPS). Although the HPWS effect on employee motivation is modified somewhat by the Great Recession, it remains robust and continues to have positive promise for small firms.

Keywords

small firms human resource management High Performance Work System workplace motivation intrinsic job satisfaction organizational commitment

For more than two decades, there has been interest within the human resource management (HRM) practitioner and research community in systems of practice that form a cohesive and integrated set designed to maximize business effectiveness and employee well-being. These systems are commonly termed high performance work systems (HPWS), or strategic human resource management (SHRM), whereby the HRM systems are tuned to harmonize with business strategic objectives. This system or strategic perspective distinguishes between HRM practices adopted by a firm in a piecemeal way and more extensive initiatives that cross several domains of people management.

Emerging evidence indicates that HPWS yield worthwhile performance gains for firms; however, most of this evidence applies to large firms. Indeed, the value of HRM/HPWS for small firms remains contentious. Some small-business experts suggest that HRM development is likely to interfere with distinctive small-firm advantages such as flexibility and informality, which small businesses should use to the full. By contrast, other experts argue that human resources repay intensive development in small firms, as these firms are often constrained in respect to other resources, notably financial.

Our focus is on the role of HRM/HPWS in the small-firm sector, which constitutes a large and growing part of the economy in the United States (Acs 1999) and in Britain (Hijzen, Upward, and Wright 2010), the country we consider here. We define “small” firms as those with fewer than 50 employees, excluding micro-businesses of 1 to 4 employees. We also show, however, that our findings hold when extended to firms with up to 100 employees. The study focuses on employee attitudes that represent motivation, using measures that have been shown in previous research to be strongly linked with individual performance.

Ours is the first British quantitative study to investigate the relationship between HRM/HPWS and employee motivation in small firms. The study makes several contributions to the debate sketched above, showing that small firms with no or minimal investment in formal HRM tend to have highly motivated employees, but that with the adoption of HRM, employee motivation declines somewhat. So far, the story accords with the critics’ warnings. In those small firms that proceed to a more intensive and integrated HRM/HPWS, however, a threshold is reached from which employee motivation climbs again. In short, the HRM-to-motivation relationship in small firms depends on the variety and complementarity of practices adopted. Positive messages as well as warnings can be drawn for small-firm practice.

A feature of our research is its coverage of two contrasting economic periods, 2004 and 2011. In 2004, economic conditions were stable and prosperous. In 2011, the British economy was struggling in the wake of a severe recession. The coverage of these two periods makes two additional contributions: first, by testing the constructive validity (Treadway et al. 2005) of HRM/HPWS effects; second, by showing, somewhat counterintuitively, that HRM/HPWS can have positive effects in the adverse context of a recession. We find that although the HRM/HPWS effect on employee motivation is modified somewhat by the Great Recession, it remains rather robust and continues to have positive promise for small firms.

Motivation and Attitudes

We view employees in small firms from a motivational perspective, through a study of their attitudes. There is a lack of consensus about the meaning of motivation and about the relationship between motivation and attitudes. Therefore, we start by outlining our notion, which comes from mainstream attitude theory and work motivation theory. We assume first that attitudes are essentially motivational. Fishbein (1967: 389) quoted with approval Louis Thurstone’s definition of an attitude (“the amount of affect for or against an object”) and equated it with a “mediating evaluative response” that tended toward overt behavior. Locke (1996: 121) continued to make values the basis of affective attitudes: “Emotions are the form in which one experiences automatized value judgements . . . according to the standard of one’s values. . . . Events and situations seen as furthering one’s values produce positive emotions (happiness, satisfaction, love).” In their review of contemporary work motivation theory, Latham and Pinder (2005) explained how attitudes express the pursuit of desired goals and values, and how the realization of goals and values sustains motivated behavior.

If this conceptual model is valid, one should observe links between attitudes and work behavior. Harrison, Newman, and Roth (2006) supplied this link with their meta-analysis that showed “overall job attitude” explains about 25% of the variation (i.e., r = 0.50) in workers’ “engagement” behaviors (task performance, organizational citizenship, attendance, timeliness, and reduced propensity to quit).

The components of overall job attitude are job satisfaction and organizational commitment. A definition of job satisfaction that fits the general framework sketched above comes from Locke and Latham (1990: 243): “The degree of satisfaction or dissatisfaction will be a joint function of the degree of fulfilment of the value and the importance of the value to the individual.” Organizational commitment then enters the frame, according to these authors, as a consequence of and complement to job satisfaction: “Only if satisfaction leads to commitment to the organization and its goals . . . will subsequent high performance result” (Locke and Latham 1990: 245). Kalleberg and Berg (1987) defined affective organizational commitment in terms of employees’ identification with the goals and values of the organization, and their willingness to exert effort on its behalf.

To represent employee motivation in the present research, we followed this approach in using measures of job satisfaction and organizational commitment but adapted the former to focus on intrinsic or autonomous satisfaction, reflecting the valued experience of autonomous working (Gagné and Deci 2005). The reason for this focus will become apparent in the following section.

Small Firms and Their Employees

We define small firms as those with fewer than 50 employees. This approach is consistent with official definitions in both the United Kingdom and the European Union, but some previous studies have used alternate definitions; for instance, Way (2002) studied US firms with 20 to 100 employees. We do not think this difference is crucial, since our variant analyses show that results are similar if we expand our sample up to firms with 99 employees.

Evidence suggests employees in small firms have particularly positive work attitudes. For instance, studies for the United States using the Quality of Employment Surveys of 1973 and 1977 reported higher satisfaction in small firms (for a review see Tansel and Gazioglu 2013 who cited 20 studies). In Britain, one can draw on studies using the Workplace Employment Relations (WERS) series, with their linked workplace and employee data in 1998, 2004, and 2011. Tansel and Gazioglu (2013) have re-analyzed the 1998 survey and reported numerous respects, including job satisfaction and perceptions of employee–management relations, in which smallness is associated with more positive attitudes. Forth, Bewley, and Bryson (2006: 41, 70), analyzing the 2004 data set, reported that small firms’ employees have the highest levels of self-rated well-being and—according to management respondents—relatively low incidence of employee grievances or disciplinary hearings. The present study is the first to use the WERS 2011 data to analyze influences on employee attitudes in small firms, but Lai, Saridakis, and Johnstone (2017) considered employee attitudes as explanatory variables for performance, and in passing (see their table 6) reported that on all the attitudinal items considered, small firms score somewhat higher than do medium-sized firms.

These findings are remarkable: bear in mind that small firms offer relatively low pay and fringe benefits, little training, and sometimes coercive forms of supervision and management (for Britain, see, e.g., Rainnie 1989; for the United States, see the extensive literature on segmented labor markets, e.g., Edwards 1979). Some insight into what seems a paradox is provided by the study of Kalleberg and Van Buren (1996). Using linked employer and employee data for the United States, they showed that larger size was significantly associated with greater material rewards but also lower feelings of job autonomy even with controls for many variables that might be linked to size. Autonomy was measured as a composite of working independently, having a say over job changes, taking part in decisions, and not being closely supervised. Interpreting this in a work motivation framework, we suggest that small firms offer greater scope for autonomous or intrinsic motivation (see especially the self-determination theory of Gagné and Deci 2005) to compensate for the relatively weak provision of extrinsic rewards. This finding also reinforces the warnings of those who have pointed to dangers in introducing HRM into small firms (see, e.g., Cardon and Stevens 2004; Marlow 2006).

British case study research provides further insight into how small firms offer intrinsic rewards. For instance, in his intensive study of three clothing manufacturers, Ram (1994) depicted an ethos of extensive freedoms and responsibilities for employees, all the more convincing because Ram’s focus was chiefly on how small businesses survived in intensely competitive markets, rather than on employees’ job quality as such. Established employees had an important role in determining their own working methods, even in one case in which the supervisor judged them to be inefficient; they socialized freely with their colleagues, and they had great discretion over their working times to fit work to the needs of their families. Their responsibilities included finding new recruits when these were needed and for training and embedding them into the work process; and, as also noted by de Kok, who examined small firms in the Netherlands (2003), employees exerted subtle but considerable influence over the decision making of their managers and principals. Case studies in restaurant businesses (Ram, Edwards, Gilman, and Arrowsmith 2001) provided a similar picture. Moule (1998), studying a small manufacturing firm, showed in detail how a work group maintained its autonomy in the face of somewhat autocratic principals.

A factor that tends to maintain a positive quality of work in some small firms is the closeness of the proprietors to employees. This aspect is strongly underlined in the Netherlands’ case studies analyzed by de Kok (2003). He observed that the employer-owner often worked alongside the employees, sought personal satisfaction in the work (as a distinct objective alongside profit), placed a high value on team spirit, and was in frequent one-to-one communication with employees, which offered scope for them to be influential.

In recent years, British small business has developed strongly in industries requiring high levels of technical and professional expertise, such as health services, information and communications technology, creative media, finance, and specialized consultancy. Case research by Tsai, Sengupta, and Edwards (2007) and by Gilman and Edwards (2008) addressed this area. The professional staff typically worked in a highly autonomous manner, with proprietors reliant on selecting people with appropriate skills. Tsai et al. (2007) reported high levels of employee satisfaction with management, in part because of the opportunities gained by having seniors work alongside them. A repeated motif in these studies is the extreme flexibility of workloads and hours, with minimal planning and a fluid type of team working that is described as “a natural extension of the way work is performed” (Gilman and Edwards 2008: 547).

Within the small-firm literature, a concept frequently deployed to describe employer–employee relationships is informality. This concept is certainly applicable to the examples of autonomous working and freedom from controls cited above, but it can also be applied to aspects of small-firm relationships that are more negative, such as proprietors’ unconcern about workplace regulations, arbitrary treatment of employees, and favoritism. At an extreme, the informal small firm can end by moving into the gray economy and the casualization of its workforce (Ram et al. 2001). Nonetheless, evidence supports that informality is valued within small firms by both owners and employees, and movement toward a more systematized or modern approach, including HRM adoption, is often resisted. In the ElectronCo case of Gilman and Edwards (2008), supervisors and team leaders were being introduced but the company was stressing that the roles would be chiefly of a mentoring rather than monitoring type. In Ram et al.’s (2001) PatCo study, a specialist food manufacturer was being pushed toward formal controls, with reduced employee discretion, as a consequence of selling to supermarkets, but management was representing the development to employees as “organized autonomy” (Ram et al. 2001: 855). Small-firm case research can be cross-checked with large-sample analysis. Storey et al. (2010) developed a survey questionnaire instrument to measure formalization and showed higher formality to be associated with lower ratings of job quality.

Our overall conclusion from this review of evidence is that small-firm employment benefits in terms of intrinsic rewards in the form of job autonomy and freedom from external controls. The strength of autonomous or intrinsic motivation (see Gagné and Deci 2005) suffices to explain the highly positive attitudes that have been reported. Much of the evidence, however, comes from small firms that have retained an informal approach, and it remains to be seen how the introduction of HRM affects the picture.

HRM/HPWS Effects on Small-Firms’ Employees

In this section, we first consider, in a general way, how HRM systems can have performance-enhancing motivational effects on employees. We then go on to discuss whether similar results can be achieved in the small-firm sector. We will focus particularly on HPWS.

Many of the leading studies in this field have used intuitive motivational concepts for interpretation and prescription. For instance, Macduffie (1995) stated that an essential condition for performance enhancement is that employees possessing knowledge and skills are motivated to apply them in discretionary effort. Economists interested in the economic effects of complementary work practices stress their value in generating incentives to productivity (Ichniowski, Shaw, and Prennushi 1997). Becker and Huselid (1998: 63) argued that the aim of HPWS was to construct a “skilled and motivated workforce providing the speed and flexibility required by new market imperatives.” Appelbaum, Bailey, Berg, and Kalleberg (2000: 46) stated that “jobs that are challenging and make use of workers’ skills are intrinsically rewarding.” Batt (2002) theorized that HPWS produced a positive effect through increased employee satisfaction that lowers the firm’s quit rate and thus helps to build up human capital and organizational learning.

Previous research (notably the seminal HPWS study of Appelbaum et al. 2000) suggested participation and team organization (team-working) were central domains of HPWS. Participation referred to methods by which employees can make contributions that directly relate to work tasks, work organization, and the management of change. Team roles supported by skill development enabled employees to widen skills, experience more challenge in their work, and experience increased relatedness with colleagues. Several more traditional aspects of HRM/personnel management have been adapted to fit into an HPWS specification (see Appelbaum et al. 2000). Financial incentives can be extended with group and/or workplace bonuses or profit-shares. Training and development can help employees take on variable job roles within teams and achieve enhanced levels of skill and self-efficacy. Recruitment and selection are complementary to training and help build a workforce committed to high performance goals (Locke 1996).

Becker and Huselid (2006), elaborating earlier contributions, argued that for HRM to have a major positive impact, relevant work practices need to be bundled in a mutually supportive way. This argument points to a threshold effect, with motivation and performance rising more steeply once the threshold has been crossed. Why this may be so is theorized more fully by Bowen and Ostroff (2004: 206), who maintain that “HRM practices can be viewed as a symbolic or signalling function.” If HRM is to alter employee behavior and performance, it must be a strong system that communicates persuasive messages: Implementing a wide range of practices is valuable in strengthening the HRM message and making it salient. This thesis connects with the idea that HPWS can project organizational values, such as developing employees’ capabilities and valuing their views, with which individuals can identify. Such a message is more likely to be trusted when the organization demonstrates its seriousness by implementing a wide range of complementary practices. Inconsistency or half-hearted dabbling in HRM, however, can be interpreted as insincerity. Although Bowen and Ostroff referred only generally to HRM, and did not specify a particular configuration of practices as ideal, a fully developed HPWS appears to meet their criteria for a strong system.

Our assumption is that the motivations of small-firm employees tend to be highly positive before any HRM development has started. As HRM practices enter the scene, they may infringe on established employee freedoms and autonomous working. Since this freedom and autonomy is the main reason for initially positive attitudes, the effect of HRM adoption is to drive motivation downward. As a firm moves closer to constructing a full HPWS, however, it signals a new participative, collaborative, and self-developing ethos with which employees can identify. To the extent that this is successful, attitudes (satisfaction and commitment) will move in a positive direction once again.

With the foregoing discussion in mind, we propose the following hypotheses:

H1: Intensive adoption of HPWS (a “strong system”) results in increased intrinsic work motivation, and this is expressed in two testable forms that we base on the work of Harrison et al. (2006) and on the more general theory of Locke and Latham (1990):

H1a: There is a positive relationship between the intensity of HPWS practices and the intrinsic job satisfaction (IJS) of employees.

H1b: There is a positive relationship between the intensity of HPWS practices and the organizational commitment (OC) of employees.

The positive relationships indicated in H1a and H1b only apply above some threshold of HPWS implementation that is to be identified empirically.

H2: At below-threshold levels of adoption of HPWS, there will tend to be reduced levels of intrinsic work motivation. This outcome will have identifiable consequences, H2a and H2b, that relate to reduction in IJS and in OC, respectively, from the high “baseline” state of motivation that we regard as characteristic of small firms.

The overall prediction, therefore, is a nonlinear (U-shaped) relationship between overall job attitudes and HPWS intensity. A model with both linear and quadratic (squared) terms represents such a relationship, with the linear term having a negative sign and the quadratic term having a positive sign.

Since our study covers two distinct economic periods or situations, the question arises whether the above hypotheses apply without modification to both. Some previous studies suggest that in firms where employee relations are adversely affected by business conditions, the positive impacts of HRM/HPWS are cancelled. For example, Zatzick and Iverson (2006) showed that during recessionary conditions in Canada, the positive link between HRM and performance was lost in firms laying off workers, whereas those firms that avoided layoffs were able to maintain this positive linkage. A British case study (Hailey, Farndale, and Truss 2005) documented the damage to employee relations when a firm overlaid a highly participative and empowering HRM system with coercive managerial behaviors responding to business pressures. By contrast, Cappelli (1999) observed that employees in the United States in the 1990s tended to be forgiving toward management that imposed layoffs, because this was seen as an action forced by external circumstances affecting most firms. In the UK recession of 2008–12, most firms were driven to make cost-reducing changes, and so management teams may have been insulated from blame by employee responses similar to those observed by Cappelli (1999). In that case, the effect of HRM/HPWS on employee attitudes and motivation would likely not be modified by the recession. Outcomes might partly depend on how baseline or pre-HRM attitudes in small firms react to recessionary pressures, but we have found no previous study directly addressing this issue. In the absence of a compelling argument to the contrary, we leave our hypotheses unaltered for the recessionary period.

Data, Measures, and Analysis Methods

Data

We use the Workplace Employment Relations Surveys for 2004 and 2011 (henceforth, WERS2004 and WERS2011).¹ We make use of employee within-firm samples to derive attitudinal outcome measures,² and management interviews (conducted prior to data collection from employees) provide the HRM/HPWS variables and other control variables. This combination reduces the risk of common method artifact (Podsakoff, MacKenzie, Lee, and Podsakoff 2003) and respects time sequencing of the independent and dependent variables (Wright, Gardner, Moynihan, and Allen 2005). In keeping with most of the literature on the effects of HRM or HPWS, we confine our analyses to the private-sector subsamples.

A sharp fall in response between 2004 and 2011 (typical of British social surveys during this period) poses a threat to comparability when we assess consistency of findings over time.³ We develop a new way of mitigating this problem, which is explained in the analysis section below.

From the management information, we identify those workplaces that represent small firms (less than 50 employees in the overall organization). Each firm is represented by only one workplace, but this is not problematic, since in the case of small firms, approximately 80% are single-establishment firms.⁴ In 2004, WERS obtained information from 280 private-sector small firms, and in 2011, from 375. The analysis focuses on these small firms.

Dependent Variables

The chief analyses refer to overall job attitude (Harrison et al. 2006) through two variables that we label intrinsic job satisfaction (IJS) and organizational commitment (OC). We obtain the measures from employee responses that are averaged at the level of the workplace that represents the firm. This aggregation and averaging process results in smooth quasi-continuous measures. Unobserved individual attributes that may bias attitudinal responses (notably personality, see Diener and Lucas 1999) will tend to be averaged out at the mean workplace level (there could, however, still be unobserved selection effects imposed by consistent selection processes). From the viewpoint of firm management, the aim is to have positive motivation across all employees, and the workplace-average measures reflect this managerial perspective.

In both years, the WERS employee questionnaire contained eight facet satisfaction items and from these, four were selected for their similarity to the “job itself intrinsic satisfaction” subscale of Warr, Cook, and Wall (1979). Table 1 provides item details. The Cronbach alpha of the IJS items in the full employee survey sample was 0.87 in both 2004 and 2011. Responses are summed at the level of the individual respondent, and the summed IJS scores are averaged over the employee respondents at each workplace.

Table 1.

Intrinsic Job Satisfaction (IJS) and Organizational Commitment (OC) Scales for Small Firms, 2004 and 2011

Scale	Range	2004			2011			Items
Scale	Range	Mean	s.d.	N	Mean	s.d.	N	Items
IJS scale	4–20	15.84	1.61	280	16.02	1.79	375	(Satisfaction with) Sense of achievement from your work, Scope for using your own initiative, Amount of influence over your job, The work itself. 5-point item response, high = more satisfied. Alpha = 0.87
OC scale	3–15	11.67	1.54	279	12.10	1.56	375	(Agreement that) Share many of the values of my organization, Feel loyal to my organization, Proud to tell people who I work for. 5-point item response, high = stronger agreement. Alpha = 0.85

Notes: Unweighted estimates. Alphas are based on the full employee sample.

The WERS measure of OC consists of three items (see Table 1), which have counterparts in the six-item Lincoln-Kalleberg measure of affective organizational commitment. OC has a Cronbach alpha of 0.85 in the full employee surveys of both years. To compute the measure, the three items were summed at the individual level and then averaged across the employees at each workplace.

The aggregation and averaging of these items requires an assumption that the items themselves constitute cardinal (interval) measures. Applied psychologists have been (implicitly) adopting the cardinality assumption since the introduction of Likert scales in the 1930s, accompanied by the method of summative ratings (Murphy and Likert 1937). More recently, economists have also tended to accept the cardinality assumption for the analysis of multi-point attitude scales. Recent examples include Powdthavee (2011) and Taylor, Jenkins, and Sacker (2011). Econometric methodological investigations that emphasize the advantages of the cardinal, or linear, assumption include Ferrer-i-Carbonell and Frijters (2004) and Riedl and Geishecker (2014).

To check the selection of items for our constructs, we performed a principal components analysis using the full employee sample on the eight satisfaction items, the three OC items, and eight further items relating to individual well-being. After varimax rotation, the IJS and OC constructs emerged as distinct components with high loadings on all their items. These results (not shown here) are available on request. Another distinct component was formed of the three items relating to satisfaction with pay received, training received, and job security: We label this extrinsic job satisfaction (EJS). We ran variant models with EJS as the dependent variable in place of IJS, as a test of the discriminant validity of the IJS construct.

Measures of HPWS Practice

Information about HRM practices comes from the WERS interview with the senior manager responsible for HRM or personnel management at the workplace. We consider only items that are descriptive of current practice and ignore any items that seek the manager’s opinion about climate, management–employee relationships, and so forth. British studies that similarly emphasize descriptive measures of HRM practice include Ramsay, Scholarios, and Harley (2000); Forth and Millward (2004); and Brown, Forde, Spencer, and Charlwood (2008); for North America, see, for example, Cappelli and Neumark (2001); Godard (2001); Wright et al. (2005); Osterman (2006); and Zatzick and Iverson (2006).

In the HRM-performance literature, all the HPWS items from a cross-sectional survey are usually aggregated into a single overall index of practices (e.g., Becker and Huselid 1998: 63).⁵ Scholars have often remarked, however, that this approach has not led to consistent, replicable measures of HRM practice, because of differences across studies in the available items. We find in the present study that although many descriptive items are available, they sometimes do not remain the same across the 2004 and 2011 surveys, and marked variation occurs in the statistical reliability of domain measures over time. We therefore introduce a new measurement approach, as follows. 1) Five domains that correspond with the HPWS concept of Appelbaum et al. (2000) are defined; these domains are participation, teams, development, recruitment, and incentives. Across these domains, we find 43 suitable item measures in 2004 and 44 in 2011. 2) We group items by domains, and the grouping is checked by reliability analysis. 3) In each firm, we count how many practices are reported to be present in each domain. If three or more items are present,⁶ we classify the firm as achieving “high” on that practice domain. 4) In each firm, we count how many domains are classified as “high,” and this number is taken as the HPWS-intensity score for that firm. This yields a six-point scale with values from 0 to 5. Among small firms, this scale has correlation 0.91 with the additive index of HPWS items in 2004, and 0.83 in 2011. The HPWS-intensity measure has a high degree of face validity with respect to the Bowen-Ostroff concept of a “strong system” HRM. Its criterion validity (with respect to employee attitudes) is demonstrated in the Results section below. Further, a firm can reach the “high” threshold in any given domain through a great variety of ways; and from the five domains, firms can select, through a number of ways, which ones they wish to develop: Thus, the HPWS-intensity score provides for uniqueness and equifinality in firm HRM/HPWS strategy (Becker and Huselid 2006) at both item and domain level. We believe that this method provides robust comparability across surveys; it is not necessary that the item pool be identical across time.

Our use of the label intensity score is consistent with practice in economics, in which (for example) the term intensive margin refers to the number of hours worked by employees. Many items are of the simple “present/absent” type, whereas others are derived by reducing a quantified banded variable (such as proportion of employees taking part in the practice, or time devoted to the practice) to dichotomous form by splitting at the median.⁷ Note also that while some items refer to fairly basic HRM practices, others can be regarded as toward the sophisticated (Cox, Zegelmeyer, and Marchington 2006) extreme: for instance, teams that select their own leader, or communication meetings that discuss staffing levels or company finances (see Appendix Table A.1). Overall, the item pools for both surveys provide a reasonable basis for assessing how far firms have progressed toward HPWS.⁸ Although there is doubtless scope for further refinement in scaling, we have held back from pursuing that because of the bias that would likely be introduced by having estimated variables among the regressors.

Table 2 gives further details of the derived domain-high and HPWS-intensity scores. High-scoring domains increased substantially between 2004 and 2011, notably in regard to participative practices and to incentives.

Table 2.

HRM Domains and HPWS Intensity Measure in Small Firms, 2004 and 2011

“High” on	2004	2011
Participation (%)	44	81
Teams (%)	42	46
Development (%)	46	55
Recruitment (%)	79	80
Incentives (%)	12	40
Mean HPWS intensity (s.e.):	2.23 (0.102)	3.02 (0.089)

Notes: “High” is a dummy variable for each domain, taking value 1 when three or more practices are implemented in that domain. All estimates are survey-weighted. For further details of items used in domain construction, see Appendix Table A.1.

Control Variables

Structural control variables, obtained from the management interview, are included in all the reported analyses. Industry is represented by 11 dummies, and controls indicate the percentage of workplace employees in higher (professional and managerial) occupations; the percentage in intermediate (administrative, technician, and craft) occupations; the percentage of female employees; the percentage (banded) of employees in non-permanent jobs; and a dummy for presence of recognized union(s). We could not include a variable subdividing size within the small (5–49 employees) segment, as this information was not available in 2004. We count the number of managers at the workplace, however, and create a dummy for those that had three or more managers—an indication of organizational complexity. We additionally include an item relating to job security guarantees made by management, which is an HRM variable considered important by Forth and Millward (2004), but did not combine with any of our HPWS domains.

Characteristics Special to Small Firms as Additional Control Variables

In all analyses, we include controls for further characteristics regarded as significant in the small-firm literature. The number of years that the business has been located at its present workplace, or at previous workplaces from which it has moved, is used as a measure of newness (see Cardon and Stevens 2004). Newness is divided into five bands approximating quintiles of the firm-age distribution. To represent family control over the firm, a frequently noted characteristic of small firms, we constructed a dummy based on whether a family holds more than half the shareholding. We also included the Storey et al. (2010) measure of formality that can be regarded as particularly relevant to small firms. Full details of this measure are given in Storey et al. (2010: 311); it has a Cronbach alpha of 0.77 in these surveys.

Firms’ Recessionary Policies in 2011

In analyzing the 2011 data, we incorporate a dummy variable identifying workplaces having more than one type of employment policy in response to the recession—commonly a wage freeze coupled with some restriction of hours or change of hours contract (e.g., zero hours contracts). One-third of small firms had two or more employment policies responding to the recession.⁹

Table 3 presents correlations between the independent variables in 2004 (panel A) and 2011 (panel B) for numeric (continuous and multi-categorical) variables followed by the tetrachoric correlations for dummy variables, excluding industry dummies.

Table 3.

Correlations of Regressor Variables

Panel 3(A) Year 2004
1. Correlations of numeric regressors
	Hi-hr	% fem	% higher	%inter	formal	age	nonperm
HI-hr	1.0
% Female	0.205	1.0
% Higher	0.159	−0.057	1.0
% Intermediate	0.069	−0.119	−0.079	1.0
Formality	0.647	0.222	0.307	0.185	1.0
Firm age	−0.164	0.005	−0.188	0.070	−0.143	1.0
% Nonpermanent	0.111	0.087	0.042	0.044	0.126	−0.122	1.0
Notes: Firm age and % nonpermanent are banded variables; as the bands are based on approximate quintiles, we treat them here as numeric but in models they are treated as categorical. HI-hr is the index of high-scoring HRM/HPWS domains.

2. Tetrachoric correlations of dummy variables
	Family	Union	Managers	Security
Family-owned	1.0
Unionized	−0.296	1.0
3+ managers	0.095	−0.067	1.0
Job security	−0.037	0.224	0.112	1.0
Notes: Matrix adjusted to be positive semi-definite. Job security means a promise by the firm not to make compulsory redundancies. Industry dummies are not included.

Panel 3(B) Year 2011
1. Correlations of numeric regressors
	Hi-hr	% fem	% higher	%inter	formal	age	nonperm
HI-hr	1.0
% Female	0.074	1.0
% Higher	0.146	−0.048	1.0
% Intermediate	−0.018	−0.205	−0.089	1.0
Formality	0.648	0.111	0.237	0.023	1.0
Firm age	−0.105	0.032	−0.085	0.101	−0.074	1.0
% Nonpermanent	0.105	0.095	−0.029	−0.011	0.018	0.030	1.0
Notes: Firm age and % nonpermanent are banded variables; as the bands are based on approximate quintiles, we treat them here as numeric but in models they are treated as categorical. HI-hr is the index of high-scoring HRM/HPWS domains.

2. Tetrachoric correlations of dummy variables
	2+ cuts	Family	Union	Managers	Security
2+ cost-cuts	1.0
Family-owned	0.040	1.0
Unionized	−1.0	−0.495	1.0
3+ managers	0.229	0.068	0.083	1.0
Job security	−0.317	−0.147	−1.00	−0.059	1.0

Notes: Matrix adjusted to be positive semi-definite. 2+ cost cuts refer to cuts made in response to the 2008 recession. Job security means a promise by the firm not to make compulsory redundancies. The perfect negative correlations between union and cuts, union and job security probably arise because of small number of unionized firms. Industry dummies are not included.

Analysis Method

The analysis uses the survey regression method with a robust variance estimator (also known as robust regression: see Berk 1990). The measures of IJS and OC are treated as continuous variables, since they are smoothly distributed workplace means. These means are themselves sample-based estimates. They are therefore measured with error and are heteroskedastic because the workplace samples vary in size. As these are always dependent variables, however, measurement error is incorporated in the usual disturbance term, and this does not affect consistency of estimates. The robust variance estimator allows for heteroskedasticity as well as for complex survey design including weighting.

The HPWS-intensity variable is represented in two different ways in variant specifications. In the first variant, it is represented as a linear effect; in the second, it is represented with both linear and quadratic (squared) terms. The latter specification makes it possible to assess the existence of nonlinearities (U-shaped relationship) as specified in the hypotheses.

The lower response rate in 2011 compared with 2004 suggests possible bias from sample selectivity. Effects on the covariance structure are unpredictable. Both surveys employed stratified sampling by workplace size and industry, and establishment weighting is intended to restore representativeness with respect to these variables. However, this does not guarantee representativeness with respect to the other control variables used in our analyses. We therefore use an alternative approach derived from the statistical matching methodology used in program evaluation research (Frölich, Huber, and Wiesenfarth 2015). We take WERS2004 as the target sample, both because of its superior response rate and because of the more typical economic conditions in which it took place, and re-weight WERS2011 to achieve mean covariate balance across all control variables that are present in both 2004 and 2011. This strategy is made possible by the entropy-balancing program developed by Hainmueller and Xu (2013). We carry out the 2011 analyses along the same lines as for 2004, but with control variables that, when the sample is re-weighted, have the same mean values (within a small tolerance) as in 2004. For example, before re-weighting, the small firms in 2011 report having a mean of 49% of employees in lower-skilled occupations, but after re-weighting this mean falls to 43%. Other variables that are substantially modified by rebalancing are the proportion of family-controlled business, the proportion of non-permanent employees, and the proportion of firms with a trade union. By reducing mean differences in control variables between surveys, we render comparative assessment more plausible. At the same time, however, we respect theoretically relevant differences between surveys by permitting the HPWS measure to vary and by introducing the additional measure of policy response to the recession in the 2011 analysis.

In variant analyses, we run the 2011 data without the re-balancing procedure and obtain HRM/HPWS effects that are neither significant nor readily interpretable. To the extent that our analyses after carrying out covariate rebalancing yield a convincing story, the method’s value is supported.¹⁰

Results

HRM/HPWS and Overall Attitudes in Small Firms, 2004

Table 4 shows the key results from regression analyses for 2004. In models (1) and (3), referring respectively to IJS and OC, the coefficient of HRM-intensity is negative and significant (albeit only at the 10% level in the case of IJS). This finding appears to support those who have argued that HRM is ill-suited to small firms’ employment relationships.

Table 4.

Robust Regression Estimates of HR Intensity on Small Firm Employees’ IJS and OC, 2004

	Mean IJS		Mean OC
	Model (1)	Model (2)	Model (3)	Model (4)
HR intensity
HI	−0.192 (0.109)	−0.983 (0.334)	−0.199 (0.096)	−0.855 (0.265)
	−1.77 +	−2.94 **	−2.07 *	−3.23 **
HI²		0.156 (0.056)		0.130 (0.048)
		2.77 **		2.73 **
Turning point		3.15		3.29
Smallness features
Formality	0.058 (0.069)	0.090 (0.072)	0.111 (0.057)	0.138 (0.059)
	0.84	1.26	1.95 +	2.33 *
Family-owned	−0.245 (0.248)	−0.240 (0.244)	−0.290 (0.213)	−0.288 (0.214)
	−0.99	−0.98	−1.36	−1.35
3+ managers	−0.070 (0.246)	−0.104 (0.242)	−0.345 (0.220)	−0.377 (0.218)
	−0.28	−0.43	−1.57	−1.73 +
Firm age, in years
7–12	−0.186 (0.339)	−0.022 (0.358)	0.093 (0.289)	0.225 (0.305)
	−0.55	−0.06	0.32	0.74
13–20	−0.018 (0.416)	0.103 (0.413)	−0.107 (0.349)	−0.007 (0.344)
	−0.04	0.25	−0.31	−0.02
21–31	0.294 (0.433)	0.438 (0.411)	0.353 (0.358)	0.470 (0.336)
	0.68	1.06	0.99	1.40
> 31	−0.576 (0.404)	−0.417 (0.400)	−0.347 (0.296)	−0.216 (0.303)
	−1.42	−1.04	−1.17	−0.71
R-squared	0.174	0.208	0.286	0.312

Notes: Cells present coefficient estimates with standard errors in parentheses and t-statistics in italics below. HI is the index of high-scoring domains (range 0–5). All above analyses have N = 276. Analyses have additional controls for industry, proportion female employees, proportions higher-level and intermediate-level employees, proportion (banded) of non-permanent employees, trade union recognition, and no compulsory redundancy policy.

Significance: + significant at the 10% level; * significant at the 5% level; ** significant t at the 1% level.

Results from models (2) and (4), which introduce a nonlinear functional form, paint a different picture. Whereas the linear term is negative in both models, the quadratic is positive, and both are statistically significant at the 1% level. The turning-point row indicates the value at which the effect of HRM intensity changes from negative to positive. For both IJS and OC, small firms can expect positive outcomes once they have three HRM domains substantially developed (approximately 40% of the small firms had reached this level of HRM development), with further improvement as they proceed toward a more complete HPWS. These results provide strong evidence in support of both H1 (a and b) and H2 (a and b).

Table 4 also reports the estimated effects of formality, family control, internal structural complexity (three or more managers), and age of firm. Somewhat unexpectedly, the level of formality is positively and significantly related to OC and is positive but nonsignificant in relation to IJS. This outcome is inconsistent with the results reported by Storey et al. (2010), who included no representation of HRM intensity. The other small-firm characteristics had little effect on the IJS and OC outcomes.

Estimates for the other control variables are not shown for reasons of space, but the full results are available on request.

HRM/HPWS and Overall Attitudes in Small Firms, 2011

Table 5 shows the results for 2011, with weighting that achieves covariate balance to 2004. This table follows the same general pattern as in Table 4, except for the inclusion of an additional variable that represents the use of multiple cost-cutting policies by the firm in response to the recession. This variable is key to understanding the 2011 results. Employees in the one-third of small firms using multiple cost-cutting methods have, other things being equal, substantially lower levels of IJS and OC. With respect to the HPWS effect, for IJS the nonlinear model continues to perform well, with the linear term significant at the 10% level and the quadratic term significant at the 5% level; IJS begins to climb once a firm has achieved, roughly speaking, substantial implementation of more than two HRM domains (by 2011 about two-thirds of small firms had reached this stage of development). In the case of OC, however, the nonlinear model fails and a simple linear model is adequate: Substantial development of any HRM domain is associated with higher OC.

Table 5.

Robust Regression Estimates of HR Intensity on Small Firm Employees’ IJS and OC, 2011

	Mean IJS		Mean OC
	Model (1)	Model (2)	Model (3)	Model (4)
HR intensity
HI	0.157 (0.113)	−0.631 (0.328)	0.187 (0.090)	−0.013 (0.276)
	1.40	−1.92 +	2.09 *	−0.05
HI²		0.136 (0.056)		0.035 (0.048)
		2.42 *		0.72
Turning point		2.32		none
Recession policies
Multiple cuts	−0.887 (0.280)	−0.912 (0.278)	−0.560 (0.256)	−0.566 (0.258)
	−3.17 **	−3.28 **	−2.18 *	−2.20 *
Smallness features
Formality	−0.162 (0.064)	−0.153 (0.062)	−0.082 (0.050)	−0.080 (0.050)
	−2.53 *	−2.49 *	−1.62	−1.59
Family-owned	−0.068 (0.253)	−0.019 (0.247)	−0.044 (0.209)	−0.031 (0.207)
	−0.27	−0.08	−0.21	−0.15
3+ managers	−0.181 (0.251)	−0.258 (0.255)	−0.026 (0.213)	−0.045 (0.211)
	−0.72	−1.01	−0.12	−0.21
Firm age, in years
7–12	−0.852 (0.518)	−0.863 (0.506)	−0.940 (0.404)	−0.943 (0.403)
	−1.64	−1.71 +	−2.33 *	−2.34 *
13–20	−0.667 (0.368)	−0.685 (0.358)	−0.738 (0.278)	−0.742 (0.278)
	−1.81 +	−1.92 +	−2.65 **	−2.67 **
21–31	−0.493 (0.388)	−0.464 (0.372)	−0.676 (0.305)	−0.669 (0.302)
	−1.27	−1.25	−2.22 *	−2.21 *
> 31	−0.831 (0.403)	−0.849 (0.395)	0.925 (0.334)	−0.929 (0.334)
	−2.06 *	−2.15 *	−2.77 **	−2.78 **
R-squared	0.184	0.201	0.174	0.175

Notes: Cells present coefficient estimates with standard errors in parentheses and t-statistics in italics below. HI is the index of high-scoring domains (range 0–5). N for these analyses is 336; the reduction in N, compared with Table 1, is mainly attributable to missing information concerning industry. For other controls, see Table 4.

Significance: + significant at the 10% level; * significant at the 5% level; ** significant at the 1% level.

To clarify what is taking place as a result of firms’ employment response to the recession, we also ran models separately for those firms that had multiple cost-cutting policy responses to the recession and those that had not (i.e., the latter comprise firms with no cost-cutting response or with only one). The estimates are shown in Table 6. When multiple cost-cutting responses are absent (panel B), the effect of HPWS intensity is similar to 2004 in the case of IJS, but disappears in the case of OC. When multiple cost-cutting responses are present (panel A), a simple linear model with positive coefficient is now supported for both IJS and OC, whereas the nonlinear model is maintained in a much weakened form for only IJS. Overall then, it seems as if the recessionary pressures transmitted to employees through small-firm cost-cutting policies in the recession tended to increase the positivity of HRM/HPWS effects (the opposite result to the Canadian study of Zatzick and Iverson 2006). A possible interpretation is that recessionary policies erase the baseline motivational advantage of small firms but not the positive impact of HPWS. Note that some caution is needed over the magnitude of point estimates in models based on 111 observations, as there is a risk of over-fitting.

Table 6.

Effects of HR Intensity in Small Firms with and without Multiple Recessionary Policies, 2011

	(A) Firms with 2+ cost-cutting policies				(B) Firms with only one or no cost-cutting policy
	Mean IJS		Mean OC		Mean IJS		Mean OC
	Model (1)	Model (2)	Model (3)	Model (4)	Model (5)	Model (6)	Model (7)	Model (8)
HR intensity
HI	0.885 (0.219)	−0.509 (0.608)	0.647 (0.189)	−0.002 (0.529)	−0.071 (0.126)	−0.803 (0.345)	−0.064 (0.083)	−0.010 (0.291)
	4.03 **	−0.84	3.42 **	−0.00	−0.56	−2.33 *	−0.77	−0.04
HI²		0.244 (0.101)		0.114 (0.093)		0.129 (0.055)		−0.009 (0.049)
		2.41 *		1.23		2.33 *		−0.19
Turning point		1.04		None		3.12		None
Smallness features
Formality	−0.256 (0.125)	−0.233 (0.115)	−0.208 (0.114)	−0.197 (0.109)	−0.095 (0.080)	−0.093 (0.076)	0.036 (0.055)	0.035 (0.055)
	−2.04 *	−2.02 *	−1.83 +	−1.80 +	−1.22	−1.23	0.65	0.65
Family-owned	−0.497 (0.474)	−0.279 (0.448)	−0.467 (0.482)	−0.365 (0.483)	−0.002 (0.270)	0.007 (0.263)	0.005 (0.203)	0.004 (0.203)
	−1.05	−0.62	−0.97	−0.76	−0.01	0.03	0.02	0.02
3+ managers	−1.035 (0.490)	−1.460 (0.488)	−0.101 (0.490)	−0.298 (0.508)	−0.027 (0.251)	−0.032 (0.250)	−0.192 (0.199)	−0.92 (0.199)
	−2.11 *	−2.99 **	−0.21	−0.59	−0.11	−0.13	−0.96	−0.96
Firm age, in years
7–12	−1.800 (0.644)	−1.944 (0.679)	−0.480 (0.598)	−0.547 (0.621)	−0.557 (0.470)	−0.540 (0.452)	−1.018 (0.338)	−1.020 (0.336)
	−2.80 **	−2.86 **	−0.80	−0.88	−1.19	−1.19	−3.01 **	−3.03 **
13–20	−1.700 (0.638)	−2.049 (0.689)	−0.855 (0.530)	−1.018 (0.572)	−0.146 (0.403)	−0.084 (0.385)	0.260 (0.269)	−0.264 (0.272)
	−2.67 **	−2.97 **	−1.61	−1.78 +	−0.36	−0.22	−0.96	−0.97
21–31	−1.297 (0.690)	−1.462 (0.734)	−0.191 (0.634)	0.268 (0.656)	−0.414 (0.402)	−0.339 (0.382)	−0.879 (0.304)	−0.884 (0.305)
	−1.88 +	−1.99 *	−0.30	−0.41	−1.03	−0.89	−2.89 **	−2.89 **
> 31	−1.871 (0.646)	−1.971 (0.718)	−0.988 (0.638)	−1.034 (0.675)	−0.606 (0.422)	−0.624 (0.406)	−0.799 (0.301)	−0.798 (0.301)
	−2.90 **	−2.74 **	−1.55	−1.53	−1.44	−1.54	−2.66 **	−2.65 **
R-squared	0.380	0.424	0.350	0.362	0.226	0.241	0.288	0.288

Notes: Cells present beta coefficients with standard errors in parentheses and t-statistics in italics below. HI is the index of high-scoring domains (range 0–5). For other controls, see Table 4. N for models 1–4 is 111; for models 5–8, N is 225.

Significance: + significant at the 10% level; * significant at the 5% level; ** significant at least at the 1% level.

Additional Tests of Robustness and Validity

We carried out several tests relating to aspects of model robustness and validity; to avoid the additional complication of rebalancing the 2011 data, these tests were confined to 2004. Appendix Table A.2 shows estimates, comparable to those in Table 4, for which the definition of small firm is extended to 5 to 99 employees. The estimates change very little with this extension, suggesting that the methodology is robust to moderate changes in population definition.

We conducted variant analyses with the HPWS intensity variable based on domains that met a criterion of two or more practices adopted (instead of three or more). The results are similar to those shown in Table 4. Confining attention to the linear-quadratic specifications, for IJS the HI effect is −1.16 with t-statistic of −2.86 and the HI-squared effect was 0.168 with t-statistic of 2.64; both test statistics are significant at the 1% level. For OC, the corresponding effects are −0.664 and 0.094, with t-statistics of −1.95 and 1.74, both significant at the 10% level. These results indicate that our method of specifying the HPWS variable permits a degree of flexibility in the domain criterion, and this contributes to its constructive validity. When we shift the domain criterion in the other direction, to four or more practices, all estimates for the linear and quadratic HPWS terms become nonsignificant. This result may be because relatively few small firms achieved the higher criterion.

We also ran an analysis with EJS as the dependent variable. Here the results for the linear HPWS term were b = −0.34, t = −1.22, and for the quadratic term b = 0.054, t = 1.21. (Full results are available on request.) This nonsignificant result indicates the discriminant validity of IJS as against EJS, hence supporting our notion that working in a small firm is intrinsically motivating.

Conclusions

The aim of this research has been to assess the effects of human resource management and the more systematic use of such practices in High Performance Work Systems on the intrinsic job satisfaction and organizational commitment of small firms’ employees, both before and after the 2008 recession in Britain. These attitudes represent dimensions of employee motivation that previous research has demonstrated to have substantial implications for individual behavior and performance.

The analyses for 2004, when the British economy was buoyant, provide strong evidence that the effects of HRM are nonlinear (U-shaped), with negative effects at low levels of HRM implementation but positive effects once more-intensive implementation has been reached. These findings are consistent with our depiction of small-firms’ employees as having somewhat positive attitudes at a baseline where HRM is undeveloped, initially reacting negatively to the introduction of HRM, but then recovering more positive attitudes as a firm moves to a more developed HPWS that fosters participation, team work, and skill development.

Results are somewhat more complex for 2011. Differences then appeared both between the effects on IJS and the effects on OC, and between small firms that had introduced multiple cost-cutting policies to counter the recession and those that had not. For the overall 2011 sample, the U-shaped relation between HRM/HPWS was maintained for IJS, but for OC a simpler linear or additive effect of HRM/HPWS now appeared best. Further analysis showed that the linear model applied for both IJS and OC in the case of the subsample of small firms that had introduced multiple cost-cutting policies. When small firms responded to the recession through cost-cutting employment policies, the former negative effect of low-intensity HRM on motivation tended to be suppressed. This result was accompanied by strongly negative reactions toward the cost-cutting employment policies themselves. A parsimonious interpretation is that the cost-cutting policies severely constrained the autonomous working and time freedoms normally enjoyed by small-firm employees, and against that frame the development of HRM/HPWS practices appeared relatively benign to employees.

The practical implications of these findings are challenging for small enterprise management. Much of our investigation accords with criticisms of HRM: attitudes are highly positive when HRM is absent. It seems unrealistic, however, to recommend staying in this never-never land. As the enterprise grows, there is a normal, possibly inevitable, movement toward more complexity, leading management to seek a more systematized approach. Such a transition is certain to be difficult, and the early stages of HRM/HPWS implementation form part of this difficulty. The key for the small firm is to press on to a more-intensive and more-integrated form of HPWS that sends stronger signals of positive intent to employees. Descriptive information for 2011 indicates that this is the direction in which many small firms are moving. The 2011 results also suggest that in turbulent competitive conditions, that may well affect small firms for the foreseeable future, HRM/HPWS will be accepted more readily by their employees. This outcome, however, requires further confirmation; qualitative research with employees of small firms adopting HRM/HPWS would be of particular value.

Footnotes

Appendix

Table A.2.

Robust Regression Estimates of HRM/HPWS Effects on Small-Firm Employees’ IJS and OC, 2004

	Mean IJS		Mean OC
	Model (1)	Model (2)	Model (3)	Model (4)
HR intensity
HI	−0.191 (−1.77+)	−0.954 (−2.98**)	−0.190 (−2.10*)	−0.763 (−2.96**)
HI²		0.151 (2.82**)	—	0.113 (2.50*)
Turning point		3.16		3.38
R-squared	0.164 N = 352	0.196 N = 352	0.268 N = 351	0.289 N = 351

Notes: “Small” is defined as 5–99 employees. Cells present coefficient estimates with t-statistics in parentheses. HI is the index of high-scoring domains (range 0–5). For controls, see Table 4.

Significance: + significant at the 10% level; * significant at the 5% level; ** significant at the 1% level.

Acknowledgements

We thank Paul Edwards for his advice and we acknowledge the Department for Business, Energy and Industrial Strategy, the Economic and Social Research Council, the Advisory, Conciliation and Arbitration Service, and the National Institute of Economic and Social Research as the originators of the 2004 and 2011 Workplace Employee Relations Survey data, and the Data Archive at the University of Essex as the distributor of the data.

For information regarding the data and/or computer programs utilized for this study, please address correspondence to the authors at a.bryson@ucl.ac.uk.

ORCID iD

Alex Bryson

Notes

References

Acs

Zoltan J.

(Ed.). 1999. Are Small Firms Important? Their Role and Impact. Boston: Kluwer Academic Publishing.

Appelbaum

Eileen

Bailey

Thomas

Berg

Peter

Kalleberg

Arne L.

2000. Manufacturing Advantage: Why High-Performance Work Systems Pay Off. Ithaca, NY: ILR Press/Cornell University Press.

Batt

Rosemary

. 2002. Managing customer services: Human resource practices, quit rates, and sales growth. Academy of Management Journal 45(3): 587–97.

Becker

Brian E.

Huselid

Mark A.

1998. High performance work systems and firm performance: A synthesis of research and managerial implications. In Ferris

Gerald R.

(Ed.), Research in Personnel and Human Resources Management, Vol. 16, pp. 53–101. Stamford, CT: JAI Press.

Becker

Brian E.

Huselid

Mark A.

2006. Strategic human resources management: Where do we go from here? Journal of Management 32(6): 898–925.

Berk

Richard A.

1990. A primer on robust regression. In Fox

John

Long

J. Scott

(Eds.), Modern Methods of Data Analysis, pp. 292–324. Newbury Park, CA: Sage Publications.

Bowen

David E.

Ostroff

Cheri

. 2004. Understanding HRM-performance linkages: The role of the “strength” of the HRM system. Academy of Management Review 29(2): 203–21.

Brown

Andrew

Forde

Christopher

Spencer

David

Charlwood.

Andy

2008. Changes in HRM and job satisfaction, 1998–2004: Evidence from the Workplace Employment Relations Survey. Human Resource Management Journal 18(3): 237–56.

Cappelli

Peter

. 1999. The New Deal at Work: Managing the Market Driven Workforce. Boston: Harvard University Press.

10.

Cappelli

Peter

Neumark

David

. 2001. Do “high performance” work practices improve establishment-level outcomes? Industrial and Labor Relations Review 54(4): 737–75.

11.

Cardon

Melissa S.

Stevens

Christopher E.

2004. Managing human resources in small organizations: What do we know? Human Resource Management Review 14: 295–323.

12.

Cox

Annette

Zegelmeyer

Stefan

Marchington

Mick

. 2006. Embedding employee involvement and participation at work. Human Resource Management Journal 16(3): 250–67.

13.

de Kok

Johannes M.

2003. Human resource management within small and medium-sized enterprises. No. 313, Tinbergen Institute Research Series. Amsterdam and Rotterdam: Tinbergen Institute.

14.

Diener

Lucas

Richard E.

1999. Personality and subjective well-being. In Daniel

Kahneman

Diener

Schwartz

David N.

(Eds.), Well-Being: The Foundations of Hedonic Psychology, pp. 213–229. New York: Russell Sage Foundation.

15.

Edwards

Richard C.

1979. Contested Terrain: The Transformation of the Workplace in the Twentieth Century. New York: Basic Books.

16.

Ferrer-i-Carbonell

Ada

Frijters

Paul

. 2004. How important is methodology for the estimates of the determinants of happiness? Economic Journal 114: 641–59.

17.

Fishbein

Martin

(Ed.). 1967. A behavior theory approach to the relations between beliefs about an object and the attitude toward the object. In Attitude Theory and Measurement, pp. 389–400. London: John Wiley & Sons.

18.

Forth

John

Millward

Neil

. 2004. High-involvement management and pay in Britain. Industrial Relations 43(1): 98–119.

19.

Forth

John

Bewley

Helen

Bryson

Alex

. 2006. Small and Medium-Sized Enterprises: Findings from the 2004 Workplace Employment Relations Survey. London: Department of Trade and Industry.

20.

Frölich

Markus

Huber

Martin

Wiesenfarth

Manuel

. 2015. The finite sample performance of semi- and nonparametric estimators for treatment effects and policy evaluation. IZA Discussion Paper No. 8756. Bonn, Germany: Institute of Labor Economics.

21.

Gagné

Marylène

Deci

Edward L.

2005. Self-determination theory and work motivation. Journal of Organizational Behavior 26(4): 331–62.

22.

Gilman

Mark W.

Edwards

Paul K.

2008. Testing a framework of the organization of small firms: Fast growth, high tech SMEs. International Small Business Journal 26(5): 531–58.

23.

Godard

John

. 2001. High performance and the transformation of work? The implications of alternative work practices for the experience and outcomes of work. Industrial and Labor Relations Review 54(4): 776–805.

24.

Hailey

Veronica H.

Farndale

Elaine

Truss

Catherine

. 2005. The HR department’s role in organisational performance. Human Resource Management Journal 15(3): 49–66.

25.

Hainmueller

Jens

Yiqing

. 2013. ebalance: A Stata package for entropy balancing. Journal of Statistical Software 54(7): 1–18.

26.

Harrison

David A.

Newman

Daniel A.

Roth

Philip L.

2006. How important are job attitudes? Meta-analytical comparisons of integrative behavioral outcomes and time sequences. Academy of Management Journal 45(2): 305–25.

27.

Hijzen

Alex

Upward

Richard

Wright

Peter W.

2010. Job creation, job destruction and the role of small firms: Firm-level evidence for the UK. Oxford Bulletin of Economics and Statistics 72(5): 621–47.

28.

Ichniowski

Casey

Shaw

Kathryn

Prennushi

Giovanna

. 1997. The effects of human resource management practices on productivity: A study of steel finishing lines. American Economic Review 87(3): 171–88.

29.

Kalleberg

Arne L.

Berg

Ivar

. 1987. Work and Industry: Structures, Markets and Processes. New York: Plenum.

30.

Kalleberg

Arne L.

Van Buren

Mark E.

1996. Is bigger better? Explaining the relationship between organization size and job rewards. American Sociological Review 61(1): 47–66.

31.

Lai

Yanqing

Saridakis

George

Johnstone

Stewart

. 2017. Human resource practices, employee attitudes and small firm performance. International Small Business Journal 35(4): 470–94.

32.

Latham

Gary P.

Pinder

Craig C.

2005. Work motivation theory and research at the dawn of the twenty-first century. Annual Review of Psychology 56: 485–516.

33.

Locke

Edwin A.

1996. Motivation through conscious goal setting. Applied and Preventive Psychology 5(5): 117–24.

34.

Locke

Edwin A.

Latham

Gary P.

1990. Work motivation and satisfaction: Light at the end of the tunnel. Psychological Science 1(4): 240–46.

35.

Macduffie

John P.

1995. Human resource bundles and manufacturing performance: Organizational logic and flexible production systems in the world auto industry. Industrial and Labor Relations Review 48: 197–221.

36.

Marlow

Susan

. 2006. Human resource management in smaller firms: A contradiction in terms? Human Resource Management Review 16: 467–77.

37.

Moule

Chris

. 1998. Regulation of work in small firms: A view from the inside. Work, Employment and Society 12(4): 635–53.

38.

Murphy

Gardner

Likert

Rensis

. 1937. Public Opinion and the Individual. New York: Harper.

39.

Osterman

Paul

. 2006. The wage effects of high performance work organization in manufacturing. Industrial and Labor Relations Review 59(2): 187–204.

40.

Podsakoff

Philip M.

MacKenzie

Scott B.

Lee

Joong-Yeon

Podsakoff

Nathan P.

2003. Common method biases in behavioral research: A critical review of the literature and recommended remedies. Journal of Applied Psychology 88(5): 879–903.

41.

Powdthavee

Nattavudh

. 2011. Anticipation, free rider problem, and adaptation to trade union: Re-examining the curious case of dissatisfied union members. Industrial and Labor Relations Review 64(5): 1000–19.

42.

Rainnie

. 1989. Industrial Relations in Small Firms: Small Isn’t Beautiful. London: Routledge.

43.

Ram

Monder

. 1994. Managing to Survive: Working Lives in Small Firms. Oxford: Blackwell.

44.

Ram

Monder

Edwards

Paul K.

Gilman

Mark W.

Arrowsmith

James

. 2001. The dynamics of informality: Employment relations in small firms and the effects of regulatory change. Work, Employment and Society 15(4): 845–61.

45.

Ramsay

Harvey

Scholarios

Dora

Harley

Bill

. 2000. Employees and high-performance work systems: Testing inside the black box. British Journal of Industrial Relations 38(4): 501–32.

46.

Riedl

Maximilian

Geishecker

Ingo

. 2014. Keep it simple: Estimation strategies for ordered response models with fixed effects. Journal of Applied Statistics 41(11). Accessed at https://doi.org/10.180/02664763.2014.909969.

47.

Storey

David J.

Saridakis

George

Sen-Gupta

Sukanya

Edwards

Paul K.

Blackburn

Robert A.

2010. Linking HR formality with employee job quality: The role of firm and workplace size. Human Resource Management 49(2): 305–29.

48.

Tansel

Aysit

Gazioglu

Saziye

. 2013. Management-employee relations, firm size and job satisfaction. IZA Discussion Paper No. 7308. Bonn, Germany: Institute of Labor Economics.

49.

Taylor

Mark P.

Jenkins

Stephen P.

Sacker

Amanda

. 2011. Financial capability and psychological health. Journal of Economic Psychology 32: 710–23.

50.

Treadway

Darren C.

Ferris

Gerald R.

Hochwarter

Wayne A.

Perrewé

Pamela L.

Witt

L. A.

Goodman

Joseph M.

2005. The role of age in the perception of politics–job performance relationship: A three-study constructive replication. Journal of Applied Psychology 90(5): 872–81.

51.

Tsai

Chib-Ju

Sengupta

Sukanya

Edwards

Paul K.

2007. When and why is small beautiful? The experience of work in the small firm. Human Relations 60(12): 1779–1807.

52.

van Wanrooy

Brigid

Bewley

Helen

Bryson

Alex

Forth

John

Freeth

Stephanie

Stokes

Lucy

Wood

Stephen

. 2013. Employment Relations in the Shadow of Recession. Basingstoke: Palgrave Macmillan.

53.

Wald

Abraham

. 1940. The fitting of straight lines if both variables are subject to error. Annals of Mathematical Statistics 11: 284–300.

54.

Warr

Peter

Cook

John

Wall

Toby

. 1979. Scales for the measurement of some work attitudes and aspects of psychological well-being. Journal of Occupational Psychology 52(2): 129–48.

55.

Way

Sean A.

2002. High performance work systems and intermediate indicators of firm performance within the US small business sector. Journal of Management 28(6): 765–85.

56.

Wright

Patrick M.

Gardner

Timothy M.

Moynihan

Lisa M.

Allen

Mathew R.

2005. The relationship between HR practices and firm performance: Examining causal order. Personnel Psychology 58(2): 409–46.

57.

Ning

Hoque

Kim

Bacon

Nicolas

Llusar

Juan C.

2015. High-performance work systems and workplace performance in small, medium-sized and large firms. Human Resource Management Journal 25(4): 408–23.

58.

Zatzick

Christopher D.

Iverson

Roderick D.

2006. High-involvement management and workforce reduction: Competitive advantage or disadvantage? Academy of Management Journal 49(5): 999–1015.