Employee Trust and Workplace Performance

We explore the relationship between employee trust of managers and workplace performance. We present a theoretical framework which serves to establish a link between employee trust and firm performance as well as to identify possible mechanisms through which the relationship may operate. We then analyse matched workplace and employee data in order to ascertain whether the average level of employee trust within the workplace influences workplace performance. We exploit the 2004 and 2011 Work Place and Employee Relations Surveys (WERS) to analyse the role of employee trust in influencing workplace performance in both pre and post recessionary periods.Our empirical findings support a positive relationship between three measures of workplace performance (financial performance, labour productivity and product or service quality) and employee trust at both points in time. We then exploit employee level data from the WERS to ascertain the determinants of employee trust as well as how trust is influenced by measures taken by employers to deal with the recent recession. Our findings suggest that restricting paid overtime and access to training potentially erode employee trust. In addition, we find that job or work reorganisation experienced at either the employee or organisation level are associated with lower employee trust.


Introduction and Background
Given the importance of identifying determinants of firm performance for understanding both economic growth and productivity at an aggregate level, it is not surprising that a vast literature exists exploring this issue focusing on a range of measures of firm performance such as financial performance (see, for example, Machin and Stewart, 1990, McNabb and Whitfield, 1998, and Munday et al., 2003 and labour productivity (see, for example, Griliches and Regev, 1995, Oulton, 1998, and Griffiths and Simpson, 2004. Many of the studies in this area focus on the role of firm level characteristics such as capital and labour inputs in determining firm performance. It is apparent that employee behaviour may influence firm level performance given that many employees have some degree of discretion with respect to how hard they work (see, for example, Brown et al., 2011, who explore the relationship between worker commitment and workplace performance). In this paper, we focus on employee trust, specifically employee trust in management, which has attracted limited interest in the economics literature. Trust can be defined as 'firm belief in the reliability, truth, or ability of someone or something' (Oxford English Dictionary, 2013). It may be the case that employee trust in the workplace influences the behaviour of employees, which in turn affects firm performance. In particular, the extent to which employees trust that their managers will treat them honestly and fairly may influence the extent to which employees engage in opportunistic behaviour or otherwise. 1 Thus, the degree of trust that employees have in their managers may impact upon firm performance.
The role of trust in the economy is being increasingly recognised in the economics literature at both the macroeconomic level, where there has been debate, for example, on the relationship between trust and economic growth (see, for example, Knack and Keefer, 1996, 3 and, more recently, Algan and Cahuc, 2010) and at the microeconomic level, such as in the context of financial decision-making (see, for example, Guiso et al., 2008, who explore the relationship between trust and stock market participation). A recent and comprehensive survey of the literature is provided by Algan and Cahuc (2013). There are an increasing number of studies in the economics literature exploring the determinants of trust at the individual level frequently using the standard trust question from the World Values Survey and the General Social Survey: Generally speaking would you say that most people can be trusted or that you can't be too careful in dealing with people? (see, for example, Alesina andLa Ferrara, 2002, andBellemare andKröger, 2007). There has also been some interest in the applied psychology and human resource management literatures, which have tended to explore the effects of specific workplace practices on employee trust. For example, Mayer and Davies (1999) explore the effects of a performance appraisal system in one particular workplace, whilst Blunsdon and Reed (2003), using Australian workplace data, find significant correlations between HR practices (such as having formalised policies and procedures) and employee trust in management. There is also some evidence that the degree of autonomy workers have over their work is associated with increased general trust, see Grund and Harbring (2009) for European evidence.
There has been less attention paid in the economics literature, however, to the role of employee trust in the workplace and its implications for firm performance. One interesting exception is La Porta et al. (1997), who explore Fukuyama's (1995) argument that high levels of trust amongst individuals serve to enhance the performance of all institutions in society including firms. They explore the effect of trust on the performance of large organisations in 40 countries. The relative success of large firms in a country is measured by the sales of the large firms relative to GNP, where a large positive effect from general trust is found. A recent contribution in the finance literature by Goergen et al. (2012) focuses on the implications of 4 intra-firm trust for firm performance and reports empirical evidence of a positive relationship.
Their measure of intra-firm trust is based on responses to 64 questions covering staff communication, profit-sharing, internal promotion, staff turnover and training. Such measures serve to capture the degree of intra-firm trust somewhat indirectly rather than employee trust per se. It is apparent that analysis of matched employee and firm level data may be a fruitful line of enquiry in order to shed further light on the relationship between firm performance and trust by exploiting more direct measures of employee trust.
This paper seeks to fill this gap in the existing literature. We begin, in Section 2, by developing a theoretical framework which establishes a link between employee trust and firm performance as well as indicating possible mechanisms through which the relationship may operate. In Section 3, we analyse matched workplace and employee data in order to explore whether employee trust influences workplace performance. To explore the robustness of our empirical findings, we exploit the 2004 and 2011 Work Place and Employee Relations Surveys (WERS) in order to analyse the role of employee trust in influencing workplace performance in both pre and post recessionary periods. Our empirical findings support a positive relationship between three measures of workplace performance (financial performance, labour productivity, and product or service quality) and employee trust. In Section 4, we exploit employee level data in order to ascertain the determinants of employee trust to shed some light on how such trust is influenced in the workplace. Section 5 concludes the paper.

Theoretical Model
In this section we seek to establish a theoretical basis for our hypothesis of a link between employee trust and firm performance and to outline possible mechanisms through which the relationship may operate. We begin by observing that each of our measures of firm performance (financial performance, labour productivity and service or product quality) can 5 be enhanced by, amongst other things, eliciting greater employee effort, engagement with training, or, willingness to adopt new processes or work-place organisation. Our theory builds on principal-agent arguments to illustrate how higher levels of employee trust in managers can help explain improvements in each of these firm performance-enhancing factors.
The principal-agent problem concerns a principal (here the manager), who wishes to incentivise the agent (here the employee) to undertake an action that is, or may appear to the agent to be, against their own best interests. We begin by outlining a typical characterization of the principal-agent problem.
Consider an agent with action set , whose choice of action affects the value of output, ( ) and their own costs, , where . Let be the principal's preferred action. Further, assume the principal is unable to observe the agent's action (there is asymmetric information), or infer it from observing output (i.e. ( ) is not one-to-one). Since action is costly to the agent and unobservable to the principal, the principal knows the agent will have an incentive to select action, .
To simplify matters, let be an n-vector of feasible values of ( ). Let be an n-vector of probabilities, with each element, (∑ ), being the probability that is observed given the agent's action is . Given that the principal employs a payment contract ( ), we construct the followinglinear in costvon Neumann-Morgenstern utility function, ( ) , for the agent, whom we assume to be risk-averse. We assume that the principal is risk-neutral, and that their objective is to design a payment contract ( ) to maximize ( ) subject to the agent's incentive compatibility constraint: and participation constraint, with reservation utility, ̅: 6 We now explore three framings of the principal-agent model to illustrate the potential channels through which trust can influence our firm performance measures: the first allows us to see how trust can be used to elicit performance-enhancing effort, the second provides insight into how trust can engender participation and co-operation or reduce costly resistance to productivity/quality enhancing change, whilst the third demonstrates how trust can influence worker identity.

Trust Eliciting Effort 2
In this section we let the elements H and L in the agent's action set represent high ( ) and low ( ) effort. We also augment the basic model outlined above to include trust. We begin under a scenario in which the agency problem yields an equilibrium with the agent choosing action L. As such it is reasonable to assume that the agent knows, through experience, the rewards and costs associated with action L. The principal, in an effort to resolve the agency problem, wishes to assure the agent that the costs to action H are no greater than and that, given this, the contract that it offers, ( ) , satisfies the constraints Eqs. (1)  Hence, even if the principal can design a feasible reward contract ( ) which satisfies Eqs.
(1) and (2), if employee trust, , is sufficiently low then it will not be possible to resolve the principal-agent problem. Consequently, since the discount parameters ( ) and ( ) are nondecreasing, higher levels of trust can increase the prospect of a given contract resolving the 7 agency dilemma, yielding effort level H and raising productivity, quality and/or financial performance.
Notice that, given the agent is risk averse, ( ) is decreasing with uncertainty.
Hence, higher levels of employee trust can act as a buffer helping to mitigate the effects of increased uncertainty (for instance during a recession), by increasing and hence ( ) ̃ ( )

Trust Inducing Engagement with Training and Re-organisation
In this section we adopt a slightly different principal-agent framework. In this case the principal can directly observe the action of the agent where H and L now refer, respectively, to high and low levels of investment/engagement in labour training, firm re-organisation or changes in working practices (or conversely low and high levels of resistance to training or re-organisation). Again, the principal's preferred action is H.
Suppose that the principal wishes to uplift worker skills and/or reconfigure working practices or the working environment so as to achieve a new, more profitable, organizational regime. For simplicity, suppose that the principal can only achieve this new regime in a future period if the agent undertakes action H (high engagement with training and/or low resistance to change) in the current period. Otherwise, the status quo prevails. Hence we have two regimes , where represents the new regime and represents the status quo.
With no asymmetry in information about the action of the agent, the principal can set a determinate reward profile for the agent ( ) where is the wage and is the working environment associated with regime . The principal's objective is therefore to design a reward profile ( ) so as to maximize ( ) subject to an optimality constraint: where is the time-adjusted value of the cost of the organizational change and/or training; an agent incentive compatibility constraint: where, is the time-adjusted cost to the agent of the organizational change and/or training; and a participation constraint, with reservation utility, ̅: However, in the absence of trust the agent may heavily discount the claims of the principal in terms of the wage and working conditions in the new regime, or anticipate a significant understatement of the direct costs to the agent of undertaking action H, i.e. the agent may base its decisions on ( ) and instead of ( ) and . 3 Hence, even if the principal can devise a feasible reward profile which satisfies Eqs.
(3)-(5), if trust is sufficiently low (i.e. and/or are sufficiently high) then the principal may not be able to find a reward profile which incentivizes the agent to opt for action H. Again, increasing employee trust increases the range of contracts which are feasible and satisfy Eqs. (4) and (5) thereby engendering the high-performance outcome for the firm.

Trust to Change Worker Identity
Finally, we consider the possibility that building employee trust can yield a change in worker identity along the lines discussed in Akerlof and Kranton (2005). In this case, the agent's utility depends on their identity where agent identity is a function of organizational practices, or more specifically in the present situation, organizational practices which influence employee trust.
To illustrate, suppose the agent can have one of two identities, . An agent with identity x (y) has an associated 'norm' under which utility is maximized, in terms of the 9 principal's preferred (non-preferred) action in the above models, with action H (L) and deviation from this action results in loss of utility. 4 If the agent's identity is x then the principal can stimulate action H at a lower wage than if agent identity is y. Replacing the cost term in Eqs.
(1) and (2) with: where is the agent's cost under action , represents the utility that the agent achieves with identity , whilst | ( ) | is a potential penalty incurred due to any divergence from the agent's 'ideal' action given they have identity . Hence investing to build employee trust to influence worker's identitychanging worker identity from type y to type xreduces the penalty associated with action H, raising . Since both effects diminish the 'net cost' term, enhancing ( ) relative to ( ) they increase the likelihood of H relative to L.
Having motivated the link between employee trust and workplace performance from a theoretical perspective, and identified potential mechanisms through which this may operate, the remainder of the paper considers whether an empirical relationship exists between trust and performance using matched employee-employer data.

Data and Methodology
In order to explore the relationship between employee trust and workplace performance from an empirical perspective, we analyse data drawn from the Workplace and Employee Relations Surveys (WERS). The aim of the WERS is to provide nationally representative data on the state of workplace relations and employment practices in Britain. We focus on data drawn from the most recent survey, namely the 2011 WERS, which is the sixth in the series.
We also explore the robustness of our findings by analysing the 2004 WERS, which relates to 10 the pre financial crisis period and, hence, allows us to explore whether the relationship between employee trust and workplace performance varies with the prevailing economic These measures of firm performance are clearly subjective and, in addition, the response rates, which are relatively consistent across 2011 and 2004, also suggest that bias exists towards responding in the average and above categories. 6 It may be the case that the three workplace performance variables are subject to measurement error (see Bertrand andMullainathan, 2001, andMcNabb, 2008). Random measurement error makes it difficult to explain variations in workplace performance, whilst if the measurement error is correlated with the explanatory variables, this leads to spurious correlation with the subjective dependent variables (Brown et al., 2011). Such issues will arguably be mitigated since the data relating to the key explanatory variables of interest are provided by employees (i.e. trust, which, as discussed in detail below, is elicited from responses to the Employee (strongly disagree) to five (strongly agree). We then match averages of the trust measures in each work place ( ̅ ̅ ̅ ̅ ) with the workplace performance information to explore how the average level of employee trust prevailing in the workplace is correlated with workplace performance. Due to the possibility of co-linearity between the four employee trust measures, they are included independently rather than simultaneously in the specification. 8 Hence, four ordered probit specifications are modelled for each of the three measures of workplace performance conditional on each alternative measure of employee trust, ̅ , and other explanatory variables, , as follows: 9 7 Furthermore, evaluations of these subjective measures of workplace performance have indicated that their ordinal properties are unaffected by such bias (see Bryson et al., 2005). In addition, comparisons of these subjective measures and objective profitability and productivity data are found to be weakly equivalent and produce similar results (Forth and McNabb, 2008). Similar evidence is reported by Wall et al. (2004), who explore the validity of subjective measures of firm performance. 8 Indeed, the pairwise correlation coefficients between the four measures of trust are all above 0.7 and are all statistically significant at the 1% level. 9 We have also used a generalised ordered probit model and we find that the general pattern of results remains.

13
̅ (6) where the unit of analysis is the workplace, w=1, …,W, (in WERS 2011 W=1,550 andin WERS 2004 W=1,432) in which the continuous latent performance of the workplace, , is observed in discrete form through a censoring mechanism: , with j outcomes and the 's are unknown parameters to be estimated. Hence, the probability that alternative j is chosen is the probability that the latent variable, , is between two boundaries and .
It is interesting to note that, as shown in the In each of the ordered probit models of workplace performance, controls in the vector include: trade union density; firm size; industry (distinguishing between: manufacturing; electricity, gas and water; construction; wholesale and retail; hotels and restaurants; transport and communication; financial services; other business services; public administration; education; health; and other community services); public sector; years in operation; the average amount of training provided to employees; the proportion of experienced staff in the largest occupational group who had training in past year; the percentage of employees using computers; whether the workplace competes at the regional (the omitted category), national or international level; the percentage of employees by occupation (distinguishing between: 14 managers and senior officials; professional; associate professional and technical; administrative and secretarial; skilled trades; caring, leisure and other personal service; sales and customer service; process, plant and machine operatives and drivers; and routine); whether the workplace operates a profit share scheme; if employees can participate in a shared ownership scheme; if the workplace implements performance related pay; and finally whether a consultative committee is thought to be "very influential" or "fairly influential" over managerial decisions which affect the workforce.  Table 1a, it is apparent that, for financial performance and labour productivity, trust is inversely related to being in the 'about average' and 'below average' categories and positively associated with being in the 'a lot better than average' and 'better than average' categories. So higher levels of employee trust (across all four measures of employee trust) appear to be positively related to workplace financial performance and labour productivity. With respect to product or service quality, employee trust is positively associated with being in the 'a lot better than average' category and inversely associated with being in the other three categories, the positive influence on the probability of reporting the highest level of this measure of workplace performance being particularly pronounced in terms of magnitude. For example, focusing on Table 1a panel c, it is evident that each alternative measure of trust, evaluated at the mean, increases the probability that product or service quality is 'a lot better than average' by approximately 8 to 9 percentage points.

Results
Turning to Table 1b,  Overall, our findings, which support the existence of a statistically significant relationship between employee trust and workplace performance, with high levels of employee trust in their managers being associated with higher levels of relative workplace performance, are consistent with our theoretical priors. Moreover, these findings are robust 11 The WERS are cross-sectional data sets, which means that we conduct separate analysis for the 2004 and 2011 surveys. A sub sample of workplaces is, however, followed across the two waves thereby allowing some panel data analysis to be conducted. Once we condition on non-missing values for the variables used in our analysis, the sub sample comprises 584 firms. In order to explore the robustness of our findings, we estimate a random effects ordered probit model and the results are consistent with the cross-sectional findings in that employee trust is positively associated with workplace performance. We further explore robustness by employing a fixed effected ordered logit estimator. The positive association remains, although, in accordance with expectations, the statistical significance of the trust variables is reduced. Modelling financial performance on the lag of the employee trust measures, to reduce the possibility of reverse causality, yields similar results.
across four different measures of employee trust and three different measures of workplace performance, as well across the 2011 and 2004 surveys. Indeed, it would appear that the influence of employee trust on workplace performance has become more important during the recession. Again, this is consistent with our theoretical priors (see Section 2.1).

Data and Methodology
Given that the findings presented in Section 3 indicate a positive relationship between employee trust and workplace performance, the natural next step is to ascertain what influences the degree of employees' trust in their managers. We therefore analyse employee level data drawn from the WERS Employee Questionnaire. We focus on the most recent WERS, i.e. the 2011 survey, since it includes a set of questions relating to whether employees were influenced by the recent recession with respect to a variety of aspects relating to their jobs. Again, in order to analyse the robustness of our findings, we explore the determinants of the four measures of employee trust (described in Section 3 above). Given that the trust measures are ordered five-point indices, we use an ordered probit specification to model each of the four measures of trust as follows: where the unit of analysis is the employee, i=1,…,N, in workplace, w=1,…,W. The continuous latent trust of the employee, , is observed in discrete form through a censoring mechanism: , with k outcomes and the 's are unknown parameters to be estimated. Standard errors are clustered at the workplace level to account for the possibility that up to 25 employees may be observed for each workplace. 12 With respect to the explanatory variables, we include a set of job and work related characteristics, , and a set of personal characteristics, . We control for the following job and work related characteristics: the natural logarithm of the individual's weekly contractual hours; the employee's workplace tenure distinguishing between less than one year (the omitted category), 1 to less than 2 years, 2 to less than 5 years, 5 to less than 10 years and 10 years or more; how much training he/she has received during the last 12 months either paid for or organised by the employer (excluding health and safety training), none (the omitted category), less than 1 day, 1 to less than 2 days, 2 to less than 5 days, 5 to less than 10 days, 10 days or more; trade union membership; and a set of dummy variables indicating which range that the individual's weekly gross pay falls into, less than £60 (the omitted category), £61-£100, £101-£130, £131-£170, £171-£220, £221-£260, £261-£310, £311-£370, £371-£430, £431-£520, £521-£650, £651-£820, £821-£1050 and £1051 or more. 13 With respect to personal characteristics, we control for gender, age, ethnicity, marital status, health status, education, number of children and religion.
Our focus on the 2011 WERS relates to the inclusion in the Employee Questionnaire of the following question: 'Did any of the following happen to you as a result of the most recent recession whilst working at this workplace? My workload increased; My job was reorganised; I was moved to another job; My wages were frozen or cut; My nonwage benefits were reduced; My contracted working hours were reduced; Access to paid overtime was 12 Our findings are robust to employing a random effects ordered probit framework. 13 The equivalent amounts are translated into annual pay in the questionnaire.
restricted; I was required to take unpaid leave; And access to training was restricted. Thus, we include a set of control variables capturing whether (as well as how) the individual reported that he/she was affected by the recent recession where these are entered into equation (7) as binary controls. 14 It is apparent from the summary statistics presented in the final column of Table 2 that 26% of employees felt that their workload had increased as a result of the recession, with 18% reporting that their work had been re-organised.
Approximately 32% reported that their wages had been frozen or cut, contrasting with only 5% reporting that their non-wage benefits had been reduced. Access to paid overtime and access to training being restricted were reported by 17% and 12% of employees, respectively.

Results
In Table 2, for brevity, we present selected results relating to the coefficients estimated in modelling , the ordered index capturing the extent to which employees agree with the statement: Managers here can be relied upon to keep their promises. Given our focus, we present the estimated coefficients related to the job and work-related characteristics. 15 It is apparent that the amount of training received by employees is positively associated with employee trust, whereas workplace tenure, hours worked and trade union membership are all inversely associated with employee trust. With respect to pay, the highest pay category is positively associated with trust with an inverse association being apparent for the middle categories relative to being in the lowest pay category. 16 14 It should be acknowledged that the variables capture the employee's perceptions regarding whether and how they were influenced by the recession, i.e. they reflect the employee's judgements regarding the perceived causation of the effects. 15 The analogous results for the other three employee trust measures are in line with those presented in Table 2 and are available on request, as are the results pertaining to the effects of the personal characteristics of the employees. With respect to personal characteristics, being male and being in poor health are consistently associated with lower levels of trust, whilst being white or Asian are associated with reporting higher levels of trust. 16 Whilst the findings here are mostly intuitive, the negative associations between trust and tenure and trust and trade union membership are a little less obvious. Indeed, in the case of trade union membership and employee trust, Bryson (2001) finds that the association can be positive or negative dependent upon factors such as the balance of power between the union and management in the workplace, the extent to which management actively encourage union membership and members' perceptions of union effectiveness.
With respect to the set of variables relating to experiences due to the economic recession, with the exception of being required to take unpaid leave, it is apparent that the estimated coefficients are all negative and generally highly statistically significant. The marginal effects relating to this set of variables are presented in Table 3 where it can be seen that the set of variables capturing whether or not employees have been influenced by the financial crisis (with the exception of having to take unpaid leave) all have a positive influence on being in the relatively low employee trust categories and a negative influence on being in the relatively high employee trust categories. Focusing firstly on , managers here can be relied upon to keep their promises, it is apparent that restricting access to paid overtime has a relatively large inverse effect on the probability of responding in the 'agree' category, at 7 percentage points, closely followed by the size of the effects of an increased workload and access to training being restricted, both at around 5 percentage points.
Moreover the effects related to these three variables are highly statistically significant. The cumulative effect of the recession variables may play an important role in influencing employee trust. Hence, in the second part of the table we present the marginal effects associated with an index of the number of recession effects reported by the employee which ranges from zero to nine. The results indicate that a higher value of the index is associated with an increased probability of reporting the lower categories of employee trust. 17 Similar results are found for and , managers here deal with employees honestly and managers here treat employees fairly, respectively, with highly significant effects also found for job re-organisation. A slightly different pattern of marginal effects is found for , managers here are sincere in attempting to understand employees' views, with negative effects found for category 5 only. The largest inverse effect on reporting category 5 'strongly 20 agree' was once again associated with restricting access to paid overtime, with highly statistically significant effects also found for restricting access to training and job reorganisation. For example, job re-organisation is associated with around a 3 percentage point lower probability that employees 'strongly agree' that managers treat employees fairly (see panel c). It is noticeable across the four measures of employee trust that being required to take unpaid leave does not appear to influence employee trust. Such a finding may reflect differing values placed on having additional time away from work related to the earnings and effort associated with being at work. 18,19 It is apparent that the changes experienced by employees due to the recession are changes experienced at the individual level. It is also interesting to explore the influence of procedures; introduction of initiatives to involve employees; and introduction of 18 We have experimented with a variety of specifications. For example, we have incorporated controls for workplace characteristics such as: workplace size; the percentage of employees dismissed over the last year; the percentage of employees made redundant over the last year; the frequency of meetings between senior managers and the whole workforce; and the number of committees of managers and employees primarily concerned with consultation (rather than negotiation). Workplace size, which is generally inversely associated with employee trust, is the only additional control to consistently exert a statistically significant influence . The pattern of results relating to the variables capturing the effects of the recent recession remains unaltered with particularly statistically significant influences found for: my workload increased; my job was re-organised; my wages were frozen or cut; access to paid overtime was restricted; and access to training was restricted. 19 In order to explore the robustness of our findings, we repeat the analysis including workplace fixed effects. The results are generally in line with those presented above. However, in line with prior expectations, the statistical significance of some of the explanatory variables is reduced.
technologically new or significantly improved product or service. In the 2011 WERS, the second and third categories were combined as follows: introduction or upgrading of new technology (including computers). Hence, seven types of organisational change were identified in the 2011 WERS as compared to eight in the 2004 WERS.
We exploit this information to explore the relationship between employee trust and organizational change by re-estimating equation (7) above replacing the variables associated with changes experienced by employees as a result of the recent recession with the organisational change variables described above. 20 The results are summarised in Tables 4 and 5 below, which present the marginal effects associated with the organisational change variables. Table 4 presents the results relating to the 2011 WERS and Table 5 presents the results relating to the 2004 WERS. As above, it may be the case that employee trust is influenced by the cumulative effects of the various types of organisational change.
Hence, in the second part of each  types of organisational change may serve to enhance employee trust. 21 In Section 2, we set out a theoretical framework to consider how employee trust can facilitate beneficial organisation practices engendering workplace performance-enhancing behaviour. The same framework can be used equally well to explain the reverse situation: practices that damage trust and reduce workplace performance. In this Section we have found evidence of both types of practice, although the results relating to the index of organisational change suggest an inverse relationship between employee trust and the number of types of organisational change introduced.

Conclusion
We have explored the relationship between employee trust and workplace performance from a theoretical and an empirical perspective. Our theoretical framework has established a link between employee trust and firm performance and has also indicated possible mechanisms It is apparent that restricting paid overtime potentially erodes employee trust, whilst requiring employees to take unpaid leave appears to have no effect on employee trust. In addition, we find that job or work reorganisation experienced at either the employee or organisational 21 If we combine the variables representing the introduction or upgrading of computers and the introduction or upgrading of other technology in the 2004 WERS, in line with the results presented in Table 5, we find positive effects associated with this type of organisational change. We present the findings associated with keeping these two categories separate in order to allow a more precise definition of the types of change and to tie in with the specific question included in the Management Questionnaire. 24 level are associated with lower employee trust. Our findings therefore highlight the importance of employee trust for workplace performance as well as shedding some light on how such trust is influenced by job and work related characteristics.