A human capital predictive model for agent performance in contact centres

How to cite this article: Jacobs, C., & Roodt, G. (2011). A human capital predictive model for agent performance in contact centres. SA Journal of Industrial Psychology/SA Tydskrif vir Bedryfsielkunde, 37(1), Art. #940, 19 pages. doi:10.4102/sajip.v37i1.940 Orientation: Currently no integrative model exists that can explain the phenomena contributing to agent performance in the South African contact centre industry.


Introduction
The contact centre and business process outsourcing (BPO) industry, established about a decade ago in South Africa, is comprised approximately of 1500 operational contact centres that employed 150 000 -175 000 agents with a further 30 000 management and support staff in 2008 (Jones, 2008).Frost and Sullivan Consulting (2009) estimated the 2007 South African BPO market revenue at $885.2 million and the global value of the sector at $220 billion.The Everest Research Institute (Witham, 2009) estimated that the growth rate of the offshore BPO market in South Africa will reach 20% -30% by 2011-2012 to eventually grow at 50% per annum which is expected to lead to 100 000 new jobs.This window of opportunity is why the government earmarked the industry for work creation (Mpahlwa, 2008).But it certainly does have its own unique challenges -specifically from a human resources and operational perspective.
Increasing technological complexity in contact centres, variations and changes in customer expectations, as well as product and knowledge intricacies, make the critical work of the agent in tightly constrained work environments very difficult (Bagnara & Marti, 2001;Kinnie, Hutchinson & Purcell, 2000).The human factor still constitutes the strategic and competitive edge in managing customer relationships as no technology can replace skilled communication, problem solving and customer focus.
The reason for the existence of modern contact centres can be found in the benefits it offers companies (Holman, 2003), which includes cost reduction of existing functions, customer service improvement and new avenues of revenue generation.A contact centre is fundamentally defined by the integration of telephonic and computer technologies that enable agents to engage in specialist operations with the work controlled by automatic systems which virtually distribute work, determine the pace thereof and monitor performance (Ellis & Taylor, 2006;Richardson & Gillipsie, 2003).Customer-employee interactions take place with the use of display screen equipment (Holman, 2003) with access to, or inputting of, information whilst facilitating inbound, outbound, blended calls or multimedia interactions (e-mail, web and text messaging).Multimedia interactions are becoming more and more prevalent (Dimension Data, 2007) as the modern contact centre is a super-user of information and communication technology systems.The industry is the benefactor of two trends namely Internet Protocol and the convergence of voice, data and Internet services over common networks and systems.Bagnara (2000) already reported in 2000 that the following human activity issues in European Union call centres had to be analysed: work organisation, training, limited career development, health, monitoring and surveillance that lead to stress and labour relations.The 2008 Global Contact Centre Benchmarking Report (Dimension Data, 2008) tracked various human capital trends over a 10-year period from 1997 to 2007; the percentage annual agent attrition rate increased by 13% from 14% to 27% and the percentage agent absenteeism rate increased by 6% from 5% to 11%.Stress responses, high absenteeism and high turnover were considered as 'normal' occurrences in modern contact centres (Bagnara & Marti, 2001).This is often associated with difficulties in personnel recruitment, management of the contact centre and staff retention whilst still meeting overall business objectives.Management actions further impact agents' job satisfaction that affect agent retention and contact centre performance (Whitt, 2006).
These trends highlight the need for empirically tested human capital models that can successfully predict overall business performance and provide a coherent multidisciplinary view that explains the phenomenon contributing to performance in a relatively new South African contact centre industry.Good models: provide causal accounts of the world, allow one to make predictive claims under certain conditions, bring conceptual coherence to a domain of science and simplify our understanding of the world.(Mouton, 2001, p. 177) Models like these are essential for enhancing decision making in the contact centre and BPO industry.
Most of the studies found in the initial literature review focused only on a particular purpose, namely on agent and/ or business performance.Examples included the impact of satisfaction on retention resulting in higher experience levels that lead to better performance; job demands-resources linked to performance; emotional intelligence linked to agent performance; and management of quantitative performance at the cost of overall performance (Aksin, Armony & Mehrotra, 2007;Dwyer & Fox, 2006;Higgs, 2004;Nel & De Villiers, 2004;Robinson & Morley, 2006).Other research mostly concentrated on elements of the interaction between operations and human resources (Aksin et al., 2007) such as the trade-off between efficiency and quality; staffing problems with learning and turnover; and staffing problems with absenteeism and random demand.
It was therefore clear from the initial literature search, that there were no comprehensive predictive model(s) that explored or attempted to explain the causality of human capital variables on agent and contact centre performance.Published research on this topic seemed to be unsystematic and lacked an integrative research approach.
The primary focus of this article was therefore to develop a theoretically derived human capital predictive model for agent performance in contact centres based on current empirical research literature.Secondary objectives included the determination of what constitutes the building blocks of a human capital contact centre predictive model of agent performance; the description of the micro level theory that constructs each building block; and to understand how these building blocks integrate to form a coherent multidisciplinary view that explains human capital phenomena contributing to agent performance.It should be noted that the multidisciplinary view incorporated perspectives from human resource management (inclusive of human capital that refers to the knowledge, skills and abilities of employees that have an economic value to the company as per Brewster, Carey, Grobler, Holland & Wärnich, 2008, p. 323), industrial and organisational psychology, operational management and financial disciplines.In view of the fact that human capital is a developing field within Human Resource Management (HRM), this research can contribute to the fields of HRM, human capital and performance management of the contact centre and BPO environment.And lastly, an objective of the study was to develop and explicate a research method appropriate for such a theoretical, model building study.
The article is structured as follows: a brief literature review is followed by the research design; the results of the literature study are detailed under 'Findings'; and the discussion features the last section of the article.

A literature review
An initial, high-level overview of the literature suggested specific themes that could be identified and that could constitute the potential building blocks of a predictive model.These were as follows.
Person-environment Fit: Person-Environment Fit (P-E fit) is defined as the compatibility between an individual employee and a work environment that occurs when the employee's characteristics are well matched (Cox, Griffiths & Rial-González, 2000, p. 37;Kristof-Brown, Zimmerman & Johnson, 2005, p. 288).Billsberry, Marsh and Moss-Jones (2004) identified five dimensions of P-E fit: 1. Person-Organisation (P-O) fit that refers to the individualorganisation value congruence 2. Person-Vocation (P-V) fit that refers to the congruence of the occupation with a person's self-concepts 3. Person-Job (P-J) fit that refers to the match between the employees knowledge, skills and abilities (KSAs) and the job-demands 4. Person-Group (P-G) fit refers to the compatibility between individuals and their work groups 5. Person-People (P-P) fit that refers to the similarity between a person's culture preferences and those preferences of others.
A mismatch between the employee and the environment in any of these dimensions could result in stress (Le Fevre, 2003).
General (noncontact centre specific) literature states that generally the degree of P-E fit relates to job satisfaction, mental well-being, physical well-being and turnover intention (Arthur, Bell, Doverspike & Villado, 2006;Yang, Che & Spector, 2008).Recruiters considered Person-Job (P-J) fit in assessing whether the candidate can do the job and the Person-Organisation (P-O) fit in order to market the candidate to the client organisation (Kristof-Brown, 2000;Werbel & Gulliland, 1999).
Job demands-resources: Bakker et al. (2003, p. 394) and Dwyer and Fox (2006, p. 128) describe the Job Demands-Resources (JD-R) model as a heuristic model that specifies how engagement and health problems may be produced by two sets of working conditions namely job resources and job demands.Engagement is the direct opposite of burnout (Simpson, 2008, p. 7).High energy, high involvement and high efficacy are characteristic of engagement, whereas exhaustion (low energy), cynicism and inefficacy are characteristic of burnout.
Job resources, according to Bakker et al. (2003, p. 395) and Dwyer and Fox (2006, p. 129), concern the extent to which the job offers resources to individual employees that contribute to retention.Job demands (Bakker et al., 2003, p. 395;Dwyer & Fox, 2006, p. 128) concerns job characteristics that potentially cause strain in that it exceeds the employee's capability to adapt; it's an energy depletion process that starts with high job demands that lead to health problems and to longer periods of absenteeism.Demerouti, Bakker, Nachreiner and Schaufeli (2001) found that job resources and demands are the predictors of work engagement.

Human resources management:
The term human resources as referred to in this section implicates employees as assets (Liu, Combs, Ketchen & Ireland, 2007, p. 504) as opposed to the term personnel that views employees as a 'cost'.The term 'human capital' refers to the knowledge, skills and abilities of employees that have an economic value to the company (Brewster, Carey, Grobler, Holland & Wärnich, 2008, p. 323).Liu et al. (2007) grouped nine key human resource management (HRM) practices together according to three key categories that impact organisational performance most based on a meta-analysis of 92 scientific investigations.These investigations include data from over 19 000 organisations of the effects on performance of HRM practices.
Social competence, with its implications on recruitment and training, was considered as critical to agent performance by management (Belt, Richardson & Webster, 2002;Hallier, 2001).Limited cross training with two skills per agent resulted in considerable performance improvements (Ahghari & Balciog, 2009;Askin et al., 2007).Agents adjusted better to the culture of the call centre and changed from just being task orientated to a balance of focusing on interpersonal needs after training (Wassenaar, 2008).Training contributed to employee satisfaction and confidence in their jobs, which enhanced their commitment to the organisation (Malhotra, Budhwar & Prowse, 2007).Frenkel, Tam, Korczynski and Shire (1998) and Houlihan (2000) contended that training was reactive because of the pressure of making targets, solving operational issues and providing the knowledge and skills to work effectively in the shortest possible time with the emphasis on narrow contextual knowledge (company products, systems and procedures) and customer-related skills.
Literature focused on incentives observed that individual goals coupled with appropriate extrinsic incentives may improve some motivational aspects (Rose & Wright, 2005) and that incentive schemes can be used as a retention initiative (Robinson, 2006).Pay for performance compensation strategies were popular in sales contact centres with an average of 20.69% of the agent's pay reported as performance based across countries (Batt, Holman & Holtgrewe, 2009).
Regarding promotion in the contact centre industry, literature reflected the occurrence of a limited use of promotion; contact centre jobs were not necessarily low quality jobs but a diverse range of jobs was offered (Batt, 2002).Visser and Rothmann (2008) suggested that an internal promotion policy might influence affective commitment in conjunction with defining a career path and a career cycle that might prove beneficial for the reduction of burnout and turnover intentions.Monotony and repetitiveness of the job content (Carrim, Basson & Coetzee, 2006;Spies, 2006) lead to unhappiness and was aggravated by limited promotion opportunities that lead to high turnover rates.
Agent competence: Agent competence, for the purposes of this study, is defined as the skills (or abilities), knowledge, attitudes, attributes and values required in performing a task within a particular context or environment; these competencies can be technical or social in nature (Grobbelaar et al., 2004, pp. 16-17).Bagnara and Marti (2001, p. 227) summarise contact centre agents' core competence as 'understanding clients' requests and finding, accessing and manipulating knowledge in the organisation and in the cognitive artefacts' as well as having knowledge about the content of interactions.
Media integration stimulated competence towards knowledge manipulation, integration and team collaboration (Bagnara & Marti, 2001); media specialisation on the other hand required the content to be the core competence as the focus shifted to service product(s) or technical support tasks (Shah & Band, 2003).The focus, however, was still on relationship building and trust building.Axtell, Parker, Holman and Totterdell (2007) found that a prerequisite for assisting customers was the capacity to take the customer's perspective.Burgers, De Ruyter, Keen and Streukens (2000) identified key customer expectations as adaptability, assurance, empathy, and the ability to deal with authority.Hampson, Junor and Barnes (2009) and Lloyd and Payne (2008)  Well-being: Henn and Barkhuizen (2009, p. 150) defined health as a 'state of complete physical, mental and social wellbeing and not merely the absence of disease or infirmity'.Mental health is defined as: a state of well-being in which every individual realises his or her own potential, can cope with the normal stresses of life, can work productively and fruitfully, and is able to make a contribution to her or his community.(Henn & Barkhuizen, 2009, p.150)The definition incorporates the absence of negative aspects such as sickness and the presence of positive factors, such as happiness.Well-being indicators are nonspecific such as live satisfaction and more specific such as job satisfaction.The notion of job satisfaction incorporates three types namely intrinsic, extrinsic and social satisfaction (Frenkel et al., 1998, pp. 969-970).
Job demands were important predictors of health problems (Bakker, Demerouti & Schaufeli, 2003;Dwyer & Fox, 2006).Dladla (2009) reported that there was a strong positive relationship between organisational climate and job satisfaction.
Contact centre agents reported musculoskeletal symptoms in the neck and/or shoulder or arm and/or hand region (Kerstin, 2005) (2000) defined turnover intention similarly as the number of agents who left in a period of time.For purposes of this study turnover intention is defined as the intention to quit one's job.Lutrin (2005) found that stress was caused by aspects such as support from outside of work, organisational factors, feelings of being undervalued, support at work and the nature of the work itself impacted on the psychological well-being and physical well-being of employees resulting in increased absenteeism and a desire to leave the organisation.Feinberg, Kim, Hokama, De Ruyter and Keen (2000) and Fulcher (2003) referred to the link between customer satisfaction and agent retention; the performance measures in most contact centres did not reward accordingly.Arthur et al. (2006), Kristof (2006) and McCulloch and Turban (2007) established that P-O fit was a predictor of turnover intention.Batt (2002) found that turnover intention was lower when high skills, agent participation in decision-making, high relative pay and employment security were empathised.Morison (2002), in her study of the effects of electronic performance monitoring on a sample of 388 contact centre agents, reported that job dissatisfaction, alongside with age, were the only variables that predicted staff turnover.

Agent performance:
The study of performance management focuses on two areas according to Strydom (2005), namely performance management on an individual level (the focus of human resource performance management) and performance management on an organisation level (the focus of organisational performance management).The South African Bureau of Standards (2008, p. 24) defined agent performance as 'individual performance to established targets and performance standards'.
Agent competence, for purposes if this study is defined as the skills (or abilities), knowledge, attitudes, attributes and values required in performing a task within a particular context or environment, these competencies can be technical or social in nature (Grobbelaar et al., 2004, pp. 16-17).Bagnara and Marti (2001, p. 227) summarised contact centre agents core competence as 'understanding clients' requests and finding, accessing and manipulating knowledge in the organisation and in the cognitive artefacts' as well as having knowledge about the content of interactions.
Du Preez (2008) specified that contact centres use statistics on the following main categories for agent and group measures: contact volume, contact handling time, adherence (measures whether agents adhere to their work schedules implying that they report to work, log in and take breaks and lunch as planned; this is impacted by sick leave, late coming, taking breaks late as a result of long calls and training), quality monitoring (measures the quality of service provided to customers by monitoring the agent work items), utilisation (measure of how much time agents spend working as a ratio of the total time available to work; agents can be idle, active, wrapping up or unavailable), turnover rate and/or attrition (percentage of agents that terminate their employment in a given period) and cost metrics.Witham (2009) mentioned that the use of detailed financial measures such as the cost per call technique was not widely used in the assessments of agent and team-level cost-effectiveness.
Wood, Holman and Stride (2006) found strong links between performance monitoring and both customer satisfaction and absence.Whitt (2006) developed a mathematical model to analyse the benefit in contact centre performance obtained from increasing agent retention that was caused by increasing agent job satisfaction; her findings indicated that the average agent performance was a function of agent experience.

Business performance:
The South African Bureau of Standards ( 2008) defined the contact centres performance as: periodic and methodical measurement of performance in accordance with defined metrics promotes personal accountability and responsibility at all levels of the operation and serves to identify, in a timely manner, strengths and weaknesses that might impact on, or compromise, the ability to meet targets.
(The South African Bureau of Standard, 2008, p. 24) Holland (2003) stated that contact centres should really be measuring the value delivered to the customer and to the business.
The aforementioned themes or building blocks were used as key words to conduct a more in-depth literature review and to structure the model building study.The next section will consider the research design under the subheadings of research approach and research method.

Research approach
A nonempirical research paradigm was adopted and a theory or model-building approach was followed with this study.This approach enabled the development of a human capital predictive model that represented the associated variables as reported in empirical studies on the prediction of agent performance in contact centres.
A conceptual model is defined as a set (or sets) of statements that represent a phenomenon as accurately as possible (Mouton, 2001, p. 177).A deductive form of theory construction was utilised whereby sets of postulates or axioms were formulated about the phenomena measured and experienced in contact centres.Theoretical propositions were deduced from the sets of postulates until a comprehensive set of theoretical propositions were developed that could be empirically tested.See diagram 1 for a graphical representation of the model building research approach.

Research method
The systematic approach followed in the collection of relevant empirical research articles described below ensures a replicable research method that is a prerequisite of the scientific method.

Location of the data and sampling procedure
The unit of analysis was quantitative, textual and/or hybrid data relating to contact centre performance related issues that were retrieved through literature searches.The sampling frame was narrowed down to include data, written in English, found in scientific journal articles, theses, dissertations and text books that relate to contact centre performance related issues limited to the period between 2000 and 2009.
This data was found primarily in e-journals located in databases such as EBSCOhost (Academic Search Premier, Business Source Premier and PsycInfo), Emerald, Gale Group (RDS Business & Management Practices), SAePublications and ScienceDirect that covers business, management and multidisciplinary subjects.EBSCOhost, through which the majority of the articles were found, is an academic and business resource boasting a powerful search engine that provides online access to full text collections of thousands of e-journals as well as more than 100 indexing and abstracting databases.Admittance to these databases was facilitated through the http://www.uj.ac.za/library web portal.The Google Scholar functionality was utilised to crossreference the aforementioned findings.

Data gathering methods
An initial scan was conducted in order to extract relevant research from the literature (Simpson, 2008).Key words used in the search included: call centres, contact centres, person-environment fit, job demands, job resources, human resource management (inclusive of selection, compensation, training, incentive compensation, internal promotion, participation programmes, flexitime, grievance procedures and employment security), agent competence, well-being, turnover and performance.The strategy was to focus and isolate data that indicated descriptions and/or correlations between variables that were utilised in formulating postulates.Only articles with the research purpose that directly related to the key words and that demonstrated linkage to performance were considered.
Articles in journals were searched based on the aforementioned keywords and retrieved via the databases mentioned under 'Location of data and sampling procedure'.Retrieval of data in dissertations and textbooks was facilitated by utilising the respective library search engines.Reference lists of articles were reviewed for additional publications and Internet searches of professional organisations delivered some results.Articles were downloaded where possible in Portable Document Format (PDF) format and saved to a communal folder.

Research procedure
Approximately 411 manuscripts were sourced by means of the mentioned key words used in the articles in order to classify, identify descriptions and/or correlations between selected variables.Only articles that reported empirical research findings on these mentioned variables were recorded for the purpose of this study.
The exact location of textbooks and journal and other articles were accurately noted and recorded.In this manner trustworthiness was ensured in that the results can be replicated by utilising the same databases and library search engines.

Data analysis
Forty-two articles were summarised in a tabular format (in Tables 2-9) detailing the study purpose, sample and setting, method and key findings (which included reference to the limitations and methodological issues) for easy replication.These tables were compiled according to the eight building blocks mentioned earlier.The listed research articles refer to empirical linkages between theoretical constructs that were analysed.Propositions that explain empirical linkages between variables were formulated, and perceived strengths and weakness of the links were discussed for further research purposes.

Presentation of the data
The retrieved data is presented in table format for each of the different building blocks.Articles are alphabetically sorted under each heading (key words) in Tables 2-9.

Findings
The review of empirical research findings pertaining to contact centres are summarised in Tables 2-9 and are discussed under the following section headings.

Person-environment fit
Table 2 summarises the one relevant article that was found.
Reiterating Table 2, McCulloch and Turban (2007) examined the value of P-O fit as a selection tool beyond cognitive ability for predicting continued length of service and performance for call centre agents.The P-O fit added significant incremental variance in predicting employee retention, but was not related to performance.Cognitive ability, on the other hand, predicted job performance but was not related to employee retention.Note: The EBSCOhost, Emerald, Gale Group, Google Scholar, SAePublications and ScienceDirect Databases were used in the items searched.

Job demands-resources
Table 3 details the four studies that were found to be relevant.
Recapping Table 3 , Bakker et al.'s (2003) and Swart's (2006) results suggested that in order to reduce or prevent absenteeism, job demands should be reduced; and in order to increase engagement and lower turnover intentions, the availability of job resources should be considered.Involvement (commitment and dedication) acted as a mediator between job resources and turnover intentions.Job resources also resulted in a higher probability when workrelated flow was experienced (Swart, 2006).Job resources' influence resulted in differential effects on performance (number of calls, call duration and customer waiting time) although demands negatively affected performance (Dwyer & Fox, 2006).One anomaly is that call duration increased as the level of certain job resources increased even whilst job demands increased.Agents who understood the wholeness of what they do in that they solve customer problems and use their abilities, spent more time with the client, handled more incoming calls and adjusted performance in the event of increased workload, role conflicts and pace.Customer service training and employee control were found to be crucial in providing coping mechanisms for demanding tasks.Feedback, task identity and task significance interacted with role demands to predict the number of calls.A sample of 477 employees working in the call centre of a Dutch telecom.
Self-reporting questionnaires and Structural equation modelling analysis were utilised.
Job demands (work pressure, computer problems, emotional demands and changes in tasks) were predictors of health problems (r = 0.57), which were related to absence as a result of sickness.Job resources (social support, supervisory coaching, performance feedback and time control) were predictors of involvement (r = 0.56) that was related to turnover intentions.Job resources had a weak negative relationship with health problems (r = -0.25),and health problems positively influenced turnover intentions (r = 0.20).Limitations included a crosssectional design that did not provide evidence for causal relationships between the variables and selfreporting.Dwyer & Fox (2006) The study tested the effects of job resources (control, training, supervisor support, task identity, task significance, skill variety and feedback) on the relationship between job demands (role demands, workload demands and work pacing demands) and indicators of performance (call duration, number of calls and customer waiting time).
A sample of 145 agents (91 and 54 respectively) from two inbound call centres in northwest Ohio.
The method included selfreporting questionnaires and hierarchical multiple regression analysis.
It was found that training moderated the relationships between role demands and performance (β = -1.27,p < 0.05), and customer waiting time and performance (β = -2.43,p < 0.05).Secondly, the three job content variables (task identity, β = 1.92, p < 0.01; task significance, β = 1.48, p < 0.05; feedback, β = 1.89, p < 0.01) had significant moderating effects on the relationship between workload demands and performance outcomes.Thirdly, control (β = -0.78,p < 0.05) and feedback (β = -2.64,p < 0.05) both interacted with work pacing demands to predict performance.A limitation was the small sample size.Lewig & Dollard (2003) Examined the emotional demands (labour) of call centre work and its relationship to job satisfaction and emotional exhaustion within the theoretical frameworks of the job demand-control model, the effort-reward imbalance model and the job demands-resources model.
A sample of 98 South Australian call centre workers from nine call centres in Adelaide Self-reporting questionnaires and Bivariate correlation analyses were utilised.
The central role of emotional labour variables was qualitatively confirmed in the experience of emotional exhaustion and satisfaction at work.Of the three components of emotional labour examined (frequency of interactions, duration of interactions and emotional dissonance) only emotional dissonance (the dissonance between felt emotions and emotional display rather being able to express positive and negative emotions) was associated with emotional exhaustion (r = 0.43, p < 0.01) and job satisfaction (r = -0.27,p < 0.01).Emotional dissonance fully mediated the relationship between positive emotional display and emotional exhaustion (β = 0.40, p < 5.001).
A limitation was the small sample size.

Swart (2006)
The objective of the study was to investigate the relationship between job characteristics (JDR), work wellness (burnout and work engagement) and workrelated flow of call centre agents.
The sample was n = 176 from a South African insurance company.
A quantitative cross-sectional design with a self-constructed instrument to measure the unique job demands and resources (JDRS).Alongside the JDRS, the Oldenburg Burnout Inventory, the Utrecht Work Engagement Scale and the Work-Related Flow Scale were used.
Job demands were found to be pressure, working conditions, workload and job security; resources were supervision, resources availability, task freedom, pay and benefits, opportunity for growth and support.Work engagement correlated positively with workrelated flow (r = 0.71) and correlated with pay and benefits (r = 0.53), opportunity for growth (r = 0.54), supervision (r = 0.47), and task freedom (r = 0.46).Limitations included the use of a cross-sectional survey design, the sample size and the fact that results were obtained solely by self-reported measures.

Human resources management practices
Three key categories impact organisational performance as per Liu et al. (2007); this framework was adopted for discussing contact centre related HRM practices.The three key categories that impacted organisational performance were competence (selection, compensation level and training), commitment (incentive compensation and internal promotion) and context (participation programmes, flexitime, grievance procedures and employment security).
The five articles that were found are summarised in Table 4 and are discussed thereafter according to the three key categories that were mentioned earlier in the article.

Competence: Recruitment and selection, training and compensation
Three relevant studies were identified regarding selection.Callaghan and Thompson (2002) observed the increased significance of social competencies within interactive service work that gave selection and training greater salience and was used by management to address the indeterminacy of labour or (as per Van den Broek, 2003) manage unionism.The recruitment process was a means to select employees who brought the skills to potentially engage in emotional labour (Townsend, 2007).Yakubovich and Lup (2006) found that when candidates were evaluated by an objective criterion, the advantage of a referral increased with the performance of his or her referrer.When job candidates self-selected into the next stage of the online application process, the referral of any agent was more likely to continue than a nonreferral, and this likelihood increased with the performance of the referrer.
One article referred to training.Townsend (2007) found that new recruits underwent a three phase training programme: • two weeks of learning computer applications that was then consolidated with two weeks of taking calls in a A survey questionnaire was administered to call centre directors or human resource managers.A least square regression analysis was utilised.
Limited use of promotion and the absence of consolidated internal labour markets in this sector were indicated.Research limitations included that the study was limited to one country with a limited number of contact centres responding.Permanent or core employees in call centres are more likely to reap the benefits of promotion opportunities (β = 2.91, p ≤ 0.05).
Promotions are effected by sophisticated selection techniques (β = 0.06, p ≤ 0.05) as call centres that rely on the information for promotion purposes.Supervision was also indicated as a tool (β = 0.09, p ≤ 0.05) when promotion is considered.Promotion activity increases as the size of the contact centre increases (β = 1.24, p ≤ 0.15).

Townsend (2007)
The purpose of the article was to consider the paradox of the extensive recruitment and training regimes in workplaces that were faced with very high levels of turnover.
The sample was one call centre of a public utilities company in Australia.
The research method was nonparticipant observation over a seven-month period coupled with ten interviews with key personnel (a qualitative approach).
It was found that the organisation was able to offset the costs of training and recruitment through internal transfers within the larger organisation.The findings are limited in that only one case study call centre was considered with resultant generalisation implications.Van den Broek (2003) This article focused on the issue of how recruitment was explicitly or implicitly designed to recruit customer service representatives who might have been antithetic to workplace trade unionism.
The sample entailed 63 interviews between 1994 and 2001 in several large call centres in the Australian telecommunications industry.
A qualitative approach was followed in that data was collected by a combination of one-hour semi-structured audio taped interviews, telephone interviews and structured written interviews.
Three processes were identified that included, (1) the use of sophisticated recruitment processes which identified those with unitarist tendencies, (2) identifying and excluding, or blacklisting, those with union backgrounds or those who previously worked in highly unionised firms and (3) applying pressure on recruits to sign individual nonunion contracts at the appointment or promotion stage.Limitations include that the sample was restricted to one country.Yakubovich & Lup (2006) Supported the claim that a referral's chances of being hired increased with the job performance of the referrer through empirical evidence.
The sample entailed a Virtual Contact Centre's database of 9008 applications for the position of sales agent that were submitted between September 2004 and February 2005.
A regression model was estimated with the recruitment process that was decomposed into objective selection, subjective selection, and self-selection detailed in five stages.
The effectiveness of referrals by the organisations staff (the referrers) as a recruitment channel varied with the type of the recruitment stage and performance of the referrer.When candidates were evaluated by an objective criterion, the advantage of a referral increased with the performance of his or her referrer; those referred by relatively high-performing workers were significantly better than the applicants who learned about the job from Internet ads (r = -0.171,p < 0.001).When job candidates self-selected into the next stage of the online application process, the referral of any agent was more likely to continue than a non-referral (r = 0.131, p < 0.01), and this likelihood increased with the performance of the referrer.On a subjective stage, the outcome was contingent on the intricacies of the recruitment process.
Limitations included generalisation from one case study to the universe of organisations that practice recruitment through referrals and the fact that contingent workers were considered rather than employees.controlled environment (assistance provided by a training partner) • two weeks learning about more advanced processes followed again by a two-week period of consolidation • new recruits were allocated to teams and entered a twelvemonth probation period with a more than 80% completion success ratio.Most of the training focused on technical skill development and product knowledge rather than emotional labour.
No relevant studies were found focusing specifically on compensation; Benner, Lewis and Omar (2007), however, reported that in 2007 the average call centre agent was paid R76 800 a year that varied by industry and type of call centre.Batt et al. (2009) reported the annual earnings of a typical agent as $11.736.Aksin et al. (2007) mentioned that many call centres included in their compensation to their agents a number-of-resolved-calls component as a result of the significance of first call resolution on overall customer satisfaction as well as on the system load.

Commitment: Incentive compensation and internal promotion
Limited research was available regarding incentives.Aksin et al. (2007), however, stated that incentives in inbound contact centres were typically based on quantitative (calls per hour, average call times, time between calls) and qualitative (content, style, adherence to policies) aspects of calls.
One study was found on promotion.Gorjup et al.'s (2008) study reported limited use of promotion in the contact centre sector.More than half of the contact centres analysed had not promoted any agents or supervisors in the previous year.It also suggested the absence of structured promotion policies.
Call centre managers use promotion more frequently if they had a higher proportion of permanent employees and incorporated sophisticated selection techniques and work monitoring practices as supervisory tools.

Context: Participation programmes, flexitime, grievance procedures and employment security
Limited research was available; other literature was therefore referred to in this section.Regarding participation programmes, Batt (2002) found that quit rates were lower (and sales growth was higher) where employee participation in decision making was at the order of the day.
In considering flexitime, Cox et al. (2000) mentioned that control over work schedules was an important factor in job design and work organisation; this was enhanced by flexitime.Other literature mentioned that flexitime could have a positive effect on workers (Narayanan & Nath, 1982) probably because the perceived control over work schedules assisted in reducing stress; it did not necessarily change behaviour (Ronen, 1981).
Grievance procedures provided a formal mechanism to change unsatisfactory work situations and resolve workplace conflict when it arose (Liu et al., 2007).It empowered employees by offering a dispute resolution valve with managers and other employees -this contributed to retention if the procedures were fair and effective.
Employment security was important for 'high commitment' service contact centres where the tasks were complex with a high level of control in the contact centre and the market volume was lower with higher value-added than compared with mass service environments (Batt, 2002;Zapf, Isic, Bechtoldt & Blau, 2003).

Agent competence
The eight relevant studies that were found are detailed in Table 5.
Reiterating Table 5, Grobbelaar et al. (2004) identified 11 competencies as being threshold competencies in order to be minimally effective.Four competencies, however, were indicated as being performance differentiators: cross-functional awareness, initiative, persuasiveness and understanding (HRM) practices.White and Roos (2005) stated the ability to listen and to be attentive to detail and information whilst yet understanding the customer's circumstances and needs as important.Agents should stay in control of the call at all times and be knowledgeable about the product, services and methodologies.
Emotional intelligence elements correlated the highest with performance and were associated frequently with the contact centre recruitment attributes according to Higgs (2004).Nel and De Villiers (2004) found that the strongest correlation with performance in the contact centre environment occurred in the cluster of self-management and the emotional competency of self-confidence in their study to determine the relationship between emotional intelligence and job performance in a contact centre environment.Moller et al. (2004) identified that a self-assertive personality type was dissatisfied with the extent of work variety in contact centres.Van der Linde's (2005) study indicated a relationship between work performance measured by the level of financial incentives of agents and five personality traits namely, (1) analytical thinking, (2) detail consciousness, (3) conscientiousness, (4) 'structuredness' and (5) work performance.Nicholls (2006) utilised the same instrument as Van der Linde (2005) and found small to moderate correlations between personality traits and performance, and moderate to high correlations between ability and performance.Ojha and Kasturi (2005) found that agents who experience high intrinsic motivation in their work and those that were team players achieve high levels of performance.

Agent well-being
Fifteen relevant studies were found and are summarised in Table 6 and discussed thereafter.Campbell (2003) indicated that customer service staff that measured high on exhaustion and cynicism experienced less job satisfaction.Exhaustion, the most important aspect of burnout, was predicted by four categories of daily hassles as per Visser and Rothmann (2009).Visser and Rothmann (2008) found that burnout had a direct effect on turnover intentions; affective commitment partially mediated the relationship between burnout and turnover intention.Fisher et al. (2007) reported neutral overall responses for job specific well-being measures, job satisfaction and job tension levels with rewards being the most negative.Emotional pressure, as a dimension of control, underpins all the factors impacting upon job satisfaction (Rose & Wright, 2005); workbased characteristics, computer-facilitated and supervisory control were found to be direct antecedents of satisfaction.
The results of Unterslak's (2009) study indicated that affective disposition and job features affected well-being.Carrim et al. (2006) found that agents with an internal locus of control experience higher extrinsic and intrinsic job satisfaction; job satisfaction was related to a positive organisational orientation whereas job dissatisfaction was related to tardiness, absenteeism and high labour turnover.
Aspects that had strong positive effect on agents' well-being (Healy & Bramble, 2003;Holman, 2003) included a high control over work methods, procedures and communication to the customer; a degree of variety; performance monitoring that was not perceived as being intense with the focus on developing the agent; supportive team leader management; supportive human resources practices; repetitive nature of the work itself; and emotional labour.According to Grebner et al. (2003), job control and task complexity and/or variety  Higgs (2004) This research detailed a study designed to explore the relationship between the emotional intelligence of contact centre agents and ratings of their performance.
A sample of 289 agents from three United Kingdom organisations was studied from inbound contact centre environments.
A quantitative crosssectional correlational design was employed using the Environmental Impact Quotient measure.
Results included a strong relationship between overall emotional intelligence as well as several elements of the model performance and individual performance.These elements include: conscientiousness (r = 0.254, significance = 0.000), emotional resilience (r = 0.186, significance = 0.001), motivation (r = 0.183, significance = 0.002) and interpersonal sensitivity (r = 0.170, significance = 0.004).Limitations included that the sample was an inbound environment with subsequent generalisation implications; the performance measures were limited in objectivity and crossorganisational consistency.The results showed that there was not a unique personality profile in respect of work satisfaction.A limitation was that the study was based on one contact centre with resultant generalisation constraints.

Nel & De Villiers (2004)
The study aimed to determine whether there was a relationship between emotional intelligence and job performance in a call centre environment.
The sample comprised 135 call centre agents from client services, sales and administrative environments in a life insurance company.

Emotional intelligence was measured by means of the Emotional Competency
Inventory that is a multirater survey instrument based on the Self-Assessment Questionnaire.
A statistically significant and positive correlation between emotional intelligence and job performance was found.The combination of emotional competencies namely emotional self-awareness, trustworthiness, selfconfidence and influence explained the greatest degree of variance in job performance in the contact centre environment as a whole (adjusted R-square = 42.8%).
Limitations include that agents' team managers (n = 31) completed the questionnaires and that the sample was insurance industry specific.

Nicholls (2006)
The purpose of the study was to determine whether personality and measures of ability predicted job performance of call centre agents.
The Small to moderate correlations were found between personality traits and performance (average handling time as a performance measure correlated as follows with: sociable r = 0.21, participative r = 0.21, persuasive r = 0.24 and average quality as a performance measure correlated as follows with: detail conscious r = 0.28, conscientious r = 0.35, structured r = 0.38, results orientated r = 0.42) and moderate to high correlations between ability (average quality as a performance measure correlated as follows with: basic checking r = 0.28, audio checking r = 0.39) and performance.
Limitations included that the study was conducted from a concurrent validity perspective as opposed to a predictive validity perspective; a relatively small sample size might have generalisation limitations if applied to the broader population.

Ojha & Kasturi (2005)
The study identified some of the attributes salient to the performance of call centre agents.
169 agents responded from two call centres, one in India and the other in the UK.
A 95-item questionnaire was developed and surveyed by conducting critical incident, behavioural event and focused group interviews in three contact centres.This was correlated with agent performance.
Agents who experienced high intrinsic motivation in their work (b = 0.202, p = 0.062) and those that were team players (b = 0.253, p = 0.016) achieved high levels of performance.Limitations included generalisation of the results that could be impacted by the fact that the sample was restricted to two contact centres and that a random subject identification method was not followed.Agents with an internal locus of control appear to experience significantly higher general, extrinsic and intrinsic job satisfaction (Chi-square = 72).
Limitations included the demographic confines and the particular organisational setting that limited the generality of the findings.Agents were as susceptible to burnout as other workers in 'burnout-prone' occupations (M = 2.86 for emotional exhaustion, M = 2.2 for depersonalisation and M = 1.29 for diminished personal accomplishment).A limitation was the small sample size.

Holman (2003)
The paper reviewed the main causes of employee stress and well being in the call centres.
Three studies by the author and his colleagues were examined.
The method of analysis was a literature review.
Employee well-being was associated with effective job design, performance monitoring that was not perceived as being intense with the focus on developing the agent, supportive management and human resources practices.Limitation: it is unclear from Holman's (2003) review of the studies whether the sample sizes encourage generalisation of the findings.

Mphuthi (2007)
The study explored possible relationships between the work design, level of anxiety and depression in South African call centres.
56 agents from a South African inbound and outbound contact centre.
Pearson correlation analysis with three questionnaires administered a biographical questionnaire, job characteristic inventory and the anxiety and depression scale.
The manner in which work was designed would have an effect in the anxiety (r = 0.34, p = 0.009) and depression levels (r = 0.33, p = 0.012) experienced by agents.Limitations included the fact that the nonexperimental and descriptive correlation design limited inferences about cause and effect in the study; the job characteristic inventory only assessed information based on six specific work design subscales, the anxiety and depression scale as a short instrument also limited the study; and lastly the small sample size had generalisation implications.Rose & Wright (2005) The study explored the factors related to control and other work-based characteristics that impact upon employee wellbeing in call centres.
173 agents partook from an insurance organisation in England.
Factor analysis and regression modelling were utilised to isolated antecedents.
Limitations included the small and relatively homogenous sample with generalisation implications; the measuring instrument for job features did not allow for enough variance to determine linearity of the relationship between job features and well-being.
Van Wyk (2005) The aim was to identify stressors in an employee assistance programme call centre.
12 agents were interviewed.A qualitative approach was followed by conducting semistructured interviews.
Four stressors were identified: job design, role ambiguity and conflict, physical health issues and communication.A limitation was the size of the sample from only one call centre.Visser & Rothmann (2009) The purpose of the study was to investigate the relationship between daily hassles and burnout and to develop a brief daily hassle diagnostic questionnaire that can be utilised within a call centre environment.
The sample size was 394 agents from a South African service and sales call centre situated in the Western Cape and Durban.
A cross sectional survey was used with and exploratory factor analysis of the data that resulted in a 6-factor model of daily hassles.Two measuring instruments were used namely the Call Centre Daily Hassle Scale (specifically developed for this study) and the Exhaustion Subscale of the Maslach Burnout Inventory (General Survey).
Exhaustion was predicted by four categories of daily hassles namely daily demands (t = 4.64, p < 0.05), demotivating work environment (t = 3.40, p < 0.05), continuous change (t = 1.97, p < 0.50) and transportation hassles (t = 2.19, p < 0.50).Daily demands were the most significant predictor of exhaustion and the demotivating work environment the second most significant.Limitations included the use of a crosssectional design that did not have the control an experimental design would have; it also did not establish causality and the self-reporting could have been biased.Visser & Rothmann (2008) The aim of this study was to investigate the relationship between six characteristics of call centre work environments, burnout, affective commitment and turnover intentions.
The sample size was 146 South African call centre agents from a financial institution in the Western Cape.
Multiple regression analysis was used to analyse the results.Questionnaires were used to collect the data and included affective commitment and turnover intention-, burnout-and turnover intention scales.
Work overload (t = 4.07, p < 0.001), lack of career opportunities (t = 3.61, p < 0.001), skill variety (t = 2.89, p < 0.005) and emotional labour (t = 2.92, p < 0.005) were the most important predictors of burnout.Limitations included the fact that the variable measurement was based on self-reporting, the salient call centre work characteristics were restricted to six and the sample size that limited generalisation of the results.
Wegge, Van Dick, Fisher, Wecking & Moltzen (2006) The aim of the study was to combine two approaches namely linking the interplay of motivational forces to the task with those linked to the social identity of employees.
Two studies from 10 different call centres with n = 211 and n = 161 call centre agents each were examined.
The relationships of objective working conditions (inbound vs. outbound work), subjective measures of motivating potential of work and organisational identification were analysed by administering two different questionnaires for each study.
In both studies it was found that objective working conditions substantially correlated with subjective measures of work motivation (MPS).
In other words higher values of MPS should be observed for employees with outbound tasks (study 1 r = 0.16, study 2 r = 0.21), more training on the job (study 1 r = 0.23, study 2 r = 0.27), and full-time contracts (study 1 r = 0.23, study 2 r = 0.18).Limitations included substantive use of self-reporting with reliance on 'one-sided' data from the call centre agents' perspective only, the cross-sectional design of both studies versus a longitudinal design, and emotional labour and emotional dissonance, which was linked to the well-being of agents, were not considered.Werner (2006) The purpose of the exploratory study was to assess the varied dysfunctions present in two call centres of the same organisation.
27 agents participated in four groups from two different call centres of the same organisation in Cape Town, South Africa.
Qualitative approach utilising focus group methodology.
Stress, caused by ineffective systems, training processes and call centre management, was the primary forerunner of dysfunctions that occur in the call centre.Limitation: the generalisation of the results was limited as a result of the fact that only one organisation was investigated.
Please see full reference list of the article, Jacobs, C., & Roodt, G. (2011).A human capital predictive model for agent performance in contact centres.SA Journal of Industrial Psychology/SA Tydskrif vir Bedryfsielkunde, 37(1), Art.#940, 19 pages.doi:10.4102/sajip.v37i1.940¹, Correlated with job satisfaction; ², Correlated with anxiety-comfort; ³, Correlated with depression-enthusiasm.http://www.sajip.co.za organisational commitment.Werner (2006) found that stress caused by ineffective systems, training processes, call centre management and exacerbated by shift work and work-life imbalances, were the primary forerunner of dysfunctions that occur in the call centre.Other dysfunctions included eating pattern disruption, depression and health consequences.Mphuthi (2007) noted that the manner in which work was designed had an effect on the anxiety and depression levels experienced by agents; autonomy and feedback correlated to some degree with anxiety and variety, and authority correlated significantly and positively with depression.Van Wyk (2005) identified job design, role ambiguity and conflict, physical health issues and communication as stressors with literature linking the stressors to job dissatisfaction, low productivity and poor motivation.Wegge et al. (2006) found that objective working conditions substantially correlated with subjective measures of work motivation.

Turnover intention
Table 7 summarises the two relevant studies that were found.
Reiterating Table 7, Schalk and Van Rijckevorsel (2007) found that contract characteristics and workplace attitudes influenced the frequency of absenteeism and intention to leave the most whilst job characteristics and personal characteristics were less important.They noted that contrary to believe job characteristics were not experienced as problematic.Spies (2006, p. 165) found that 'emotional exhaustion affects intents to leave indirectly through organisational commitment'.The intensity of emotional display was significantly related to emotional exhaustion and affective commitment respectively; the variety of emotional display was related to supervisor support.Her findings also suggested that supervisor support plays a role in contributing to agents' organisational commitment.

Agent performance
Three studies were found to be relevant; these are detailed in Table 8 and discussed thereafter.Bain et al. (2002) found that target setting comprised quantitative measures of performance (number of calls answered, average handling time and time between calls) but also qualitative factors such as the call content, rapport with the customer during the interaction and pride in the company.Qualitative approaches were still rooted in Taylorist The purpose of the study was to examine the factors that influence frequency of absenteeism and the intention to leave in a call centre.
89 call centre employees of an insurance company with the job title 'damage claim advisor'.
A 5-point scale questionnaire was administered.Job characteristics were measured using a 12-item Dutch version of the Job Diagnostic Survey.

Spies (2006)
The purpose of the study was to determine to which extend emotional labour influences emotional exhaustion and whether the results (burnout and turnover intention) were detrimental to the organisations' success.
A sample of 84 agents from one South African contact centre.
Qualitative content analysis was conducted to analyse the focus group interviews followed by a Pearson product-moment correlation analysis and multiple regression analysis.
Limitations included a small sample size, the scale used to measure emotional labour was not refined, all the components of burnout were not included and supervisor support was not clearly defined.The purpose of the study was to consider managerial utilisation of targets in order to manage agents' quantitative and qualitative performance.
Four contact centres in Scotland.
A qualitative approach was followed by observing and facilitating semistructured and structured interviews with agents, supervisors and managers.
Call centre performance occurred within a Tayloristic approach and technological development.Limitations included the fact that the sample size consisted of only four contact centres with generalisation implications.Holman, Chissick, & Totterdell (2002) The study investigated the relationship between performance monitoring and well-being of agents.
347 customer service agents partook from two United Kingdom call centres.
Qualitative interviews were conducted to design a questionnaire that was administered on site.
The findings indicated that performance monitoring in call centres was an important antecedent of well-being and emotional labour although the impact of performance monitoring was not uniform.Limitations included the crosssectional nature resulting in the fact that the direction of causality could not be determined; common method variance could have present invalidity of substantive interpretation.Mahesh & Kasturi (2006) The study aimed to investigate the relationship between important aspects of the call centre agents' job and the agents' effectiveness.
113 agents partook in the qualitative aspect of the study and 169 from two contact centres (one in India and the other in the United Kingdom) were part of the quantitative aspect of the study.
Qualitative data was gathering through critical incidents, behavioural events interviews and focus group interviews.A questionnaire was drawn up and administered based on the findings.
Four distinct factors were identified namely intrinsic motivation, reward and/or recognition, customer stress and stress management.
A limitation is the small sample size with generalisation implications.
techniques; tasks were segmented and were supervised by means of numerical performance on a continuous basis.Holman et al. (2002) indicated that the performancerelated content (immediacy of feedback and the clarity of performance criteria) and the beneficial-purpose (focus on developmental rather than corrective action) of monitoring was positively related to well-being; the perceived intensity had a strong negative association with well-being.Job control and supervisory support did moderate the relationship between perceived intensity, and well-being and perceived intensity were strongly associated with emotional exhaustion.Job control and supervisory support were strongly associated with depression and job satisfaction.Mahesh and Kasturi (2006) found that intrinsic motivation correlated positively with effectiveness (more specifically so amongst experienced agents).Customer stress correlated negatively with intrinsic motivation and positively with reward and/ or recognition; stress management correlated positively with intrinsic motivation.This implied that in order to improve agents' performance, intrinsic motivation should be tapped into rather than control and that higher levels of intrinsic motivation could implicate a less stressed workforce.

Business performance
Four relevant studies are summarised in Table 9 and discussed thereafter.
Devenish-Meares' ( 2003) study indicated that sales performance results were positively related to strategic orientations that were more externally focused (as opposed to looking inward), psychosocial climate (in general), and employee reward and morale specifically.Feinberg et al.'s (2000) results showed that only first call resolution and abandonment rate had an influence on caller satisfaction.The purpose of the study was to investigate to which extent strategic and behavioural aspects were aligned with telephone-banking contact centres' operational-service and sales-performance objectives.
The sample size was 24 financial institutions inclusive of retail banks, building societies and credit unions in Australia.
An exploratory approach was followed in order to establish and interpret key channel performance relationships; a mail survey was facilitated that covered strategic and behavioural orientations and performance outcomes.
The results indicated that sales performance was positively related to strategic orientations (r = 0.327, p = 0.086) that was externally focused, psychosocial climate and employee reward and morale specifically.Limitations included a representative sample that did not represent the smaller credit unions, the cross-sectional nature and a limited sample size.Feinberg, Kim, Hokama, De Ruyter & Keen (2000) The study undertook an empirical assessment of the relationship between caller satisfaction and 13 critical operational variables.A quantitative approach was followed with data collected by means of a questionnaire; this was followed by a qualitative in-depth semistructured interview.
There was conflict in the strategic intent of call centres from an organisational and managerial perspective (no correlations indicated).Existing key performance indicators were rated in order of emphasis: the level of service was rated highest, sales second highest, costs third highest with customer focus fourth highest.Key performance indicators were also ranked in order of importance; customer focus was rated as the most preferred key performance indicators, and then in order of declining importance: level of service, staff turnover rate, abandonment rate, wrap-up time, call duration or average handling time, occupancy rate and lastly number of calls per agent.A limitation was the length of the questionnaire that was designed to be completed in 15 minutes in order to motivate a high response rate.Strydom (2005) The study aimed to review existing theory and literature relevant to call centre performance management performance systems; and to assess the current situation.
The sample included two interviews with representatives of a South African financial institution, 19 questionnaire respondents and three 'tool' respondents respectively.
The nature of this research was mostly exploratory (quantitative) with two assessment instruments that were used.
Findings included that current expectations did not match the required purposes of performance systems.Limitations included the small sample size that restricts the statistical assessment of the reliability and validity of the results; representativeness of performance management systems was also limited.This implied that resolving the problem or answering the question the first time and making certain that the customer gets through the first time were essential to the operation of the contact centre.Robinson and Morley (2006) found that there was conflict in the strategic intent of call centres from an organisational and managerial perspective.Organisations used contact centres as a means of reducing costs with customer service delivery as a secondary consideration.Call centre managers considered customer service as their primary responsibility.Strydom (2005) found that important purposes of performance management systems (PMS) in South Africa were to determine staff incentives and for development purposes rather than determining the profitability of the contact centre.Call centres that formed part of bigger organisations considered performance management in order to illustrate 'value creation' to its shareholders significantly more than outsourced contact centres.

Discussion
This study was motivated by the need for human capital models that can successfully predict overall business performance and provide a coherent multidisciplinary view that explains the phenomena that contribute to performance in a relatively new South African contact centre industry.The primary focus of this article was to develop a theoretically derived human capital predictive model for agent performance in contact centres based on current empirical research literature.A secondary aim of the study was to develop and explicate a research method appropriate for such a theoretical, model building study.Models and theories are very important to scientific progress (Mouton, 2001).This study added value through developing a conceptual human capital predictive model to predict agent performance in contact centres.More specifically, this study identified key variables that are empirically related to agent performance.The study also shows possible interaction effects between these variables and their relation to agent performance.The practical application in contact centres is the ability to predict the effect of human capital variables on agent performance.The discussion synthesises linkages in literature between theoretical concepts and explains causal links between propositions.

Summary of findings
The summary of findings reiterates the results and relates it back to findings of other researchers.Propositions are stated after each building block based on linkages between theoretical concepts and a diagrammatical depiction of the predictive model concludes the summary of findings.

Person-environment fit
The P-O fit dimension of the P-E fit had a negative relationship with turnover and a limited relationship with performance (McCulloch & Turban, 2007).
The mentioned finding was reiterated by noncontact centre related literature that found that the P-E fit had a positive relationship with agent well-being (Arthur et al., 2006;Yang et al., 2008) and not necessarily with performance.The P-J fit dimension of the P-E fit had a positive relationship with the selection component of HRM practices and agent competence (Kristof-Brown, 2000;Werbel & Gulliland, 1999) and should be included alongside cognitive ability in the assessment for agent performance.
The following propositions are proposed based on the aforementioned statements: • Propostion 1: The P-O fit will have a negative relationship with turnover.• Propostion 2: The P-E fit will have a positive relationship with agent well-being.• Propostion 3: The P-E fit will have a positive relationship with HRM practices.
• Propostion 4: P-E fit will have a relationship with JD-R.Dwyer and Fox's (2006) model linked job demands to performance with job resources as moderators.Bakker et al. (2003) and Swart (2006) suggested that if job demands are reduced, absenteeism will be reduced which should impact agent performance positively.

Job demands-resources
Noncontact centre related research affirmed the importance of availing job resources that increased engagement (Demerouti et al., 2001) with the subsequent high energy, high involvement and high efficacy that characterises engagement which in turn influences agent performance positively.Involvement acted as a mediator between job resources and turnover intentions.Workforce management (the capability to determine the right number of resources at the right time to respond to customer contacts in order to meet service levels whilst minimising costs) determines workload and work pace demands, and should be considered as a job-demand in future research (Du Preez, 2008).
The following propositions are proposed: • Propositions 5: There will be a negative relationship between job-demands and job performance.• Propositions 6: There will be a negative relationship between job-demands and well-being.• Propositions 7: There will be a positive relationship between job resources and job performance.• Propositions 8: Job resources will moderate the effect of job demands on work engagement.• Propositions 9: The relationship between job resources and job performance will be mediated by work engagement.• Propositions 10: Work engagement will mediate the relationship between job-demands and turnover intention.• Propositions 11: There will be a relationship between JD-R and HRM Practices.

Human resources management practices
The three key categories that impact organisational performance as per Liu et al. (2007)

Competence:
The increased significance of social competencies within interactive service work gave selection and training greater salience (Callaghan & Thompson, 2002;Van den Broek, 2003) with referrals having an advantage in the selection process (Yakubovich & Lup, 2006).South Africa's annual agent earnings were reported as $11.736 (Batt et al., 2009).
Other literature affirmed that management considered social competence as critical to agent performance (Belt, Richardson & Webster, 2002;Hallier, 2001) but counter intuitively it was found that most of the training focused on technical skill development and product knowledge (Frenkel et al., 1998;Houlihan, 2000;Townsend, 2007) as a result of performance pressures.Limited cross-training with two skills per agent had resulted in performance improvements (Ahghari et al., 2009;Aksin et al., 2007).Training furthermore assisted in adjustments (Wassenaar, 2008) and contributed to employee satisfaction (Malhotra et al., 2007).
This was affirmed by other literature that observed that individual goals coupled with appropriate extrinsic incentives improved some motivational aspects (Rose & Wright, 2005); incentive schemes could also be used as a retention initiative (Robinson, 2006).Pay for performance compensation strategies were popular in sales contact centres as a motivator for performance (Batt et al., 2009).
Limited use of promotion in the contact centre sector was reported (Batt, 2002;Gorjup, et al., 2008).
Other literature reiterated the aforementioned in that internal promotion policies were encouraged as it might influence affective commitment (Visser & Rothmann, 2008) and help to address monotony and repetitiveness of the job content (Carrim et al., 2006;Spies, 2006).
In considering HRM practices, the following propositions are proposed: • Proposition 12: HRM practices will have a positive relationship with agent competence.• Proposition 13: HRM Practices will have a positive relationship with agent well-being.• Proposition 14: HRM practices will have a positive relationship with agent performance.
• Proposition 15: HRM practices will have a positive relationship with organisational performance.• Proposition 16: The degree in which HRM practices are horizontally integrated (across HRM functions) will be positively related to organisational performance.• Proposition 17: The degree in which HRM practices that are vertically integrated (aligned with business strategy) across organisational levels will be positively related to organisational performance.

Agent competence
Various competencies were identified for minimal effectiveness with cross-functional awareness, initiative, persuasiveness and understanding practices, ability to listen and to be attentive to detail and information as performance differentiators (Grobbelaar et al., 2004;White & Roos, 2005).
Most of these competencies underpin social competence.
Other literature reiterated that media integration required knowledge manipulation, integration competence and team collaboration competencies (Bagnara & Marti, 2001;Hampson et al., 2009;Lloyd & Payne, 2008).Media specialisation on the other hand demanded content (service products or technical support tasks) to be the core competence (Shah & Band, 2003).Customer perspective taking and expectations were also found to be important competencies (Axtell et al., 2007;Burgers et al., 2000).
The following proposition is proposed: • Proposition 18: Agent competence will relate positively with agent performance.
Other literature highlighted different antecedents from the aforementioned namely: job demands were important predictors of health problems (Bakker et al., 2006;Dwyer & Fox, 2006), organisational climate was positively related to job satisfaction (Dladla, 2009), and agents reported musculoskeletal symptoms in the neck and/or shoulder or arm and/or hand region as a result of the discomfort of the work environment (Kerstin, 2005).
The following propositions are proposed: • Proposition 19: Agent well-being is positively related to agent performance.• Proposition 20: Agent well-being is inversely related to turnover intention.

Turnover intention
The frequency of absenteeism and turnover intention were influenced the most by contract characteristics and workplace attitudes (Schalk & Van Rijckevorsel, 2007).
Emotional exhaustion was causally related to the intention to leave through organisational commitment (Spies, 2006).
Other literature reported different causes of turnover intention: stress and the intention to leave were caused by dissatisfaction (Batt, 2002;Lutrin, 2005;Morison, 2002); customer satisfaction and agent retention were linked (Feinberg et al., 2000;Fulcher, 2003) with performance measures that mostly did not reward accordingly; and P-O fit was a predictor of turnover intention (Arthur et al., 2006;Kristof, 2006;McCulloch & Turban, 2007).
The following proposition is proposed: • Proposition 21: Turnover intention is negatively related to agent performance.

Agent performance
As mentioned, the P-O dimension of P-E fit, JD-R, HRM practices, agent competence, agent well-being and turnover intention all relate to individual agent performance.The performance-related content and the beneficial-purpose of monitoring were positively related to well-being (Holman et al., 2002).Intrinsic motivation correlated positively with effectiveness (Mahesh & Kasturi, 2006).
Other literature only specified that performance monitoring and both customer satisfaction and absence were linked (Wood et al., 2006) with little reference to the importance of benefits and motivation of performance.The average agent performance was more a function of agent experience (Whitt, 2006) than performance monitoring whereby contact centres utilised various statistics inclusive of cost per call metrics (Du Preez, 2008;Witham, 2009).
The following proposition is proposed: • Proposition 22: Agent performance will have a positive relationship with overall business performance.

Overall business performance
Agent performances to a large extent contributed to the overall performance of the contact centre.Sales performance results were positively related to strategic orientations that were externally focused (Devenish-Meares, 2003).Only first call resolution and abandonment rate had an influence on caller satisfaction (Feinberg et al., 2000).Conflict was found in the strategic intent of call centres from organisational and managerial perspectives (Robinson & Morley, 2006).Performance management systems were utilised for staff incentives and for development purposes rather than profitability management (Strydom, 2005).
Other literature reiterated these findings in that contact centres should really be measuring the value delivered to the customer and to the business (Holland, 2003).Figure 2 diagrammatically depicts the eight building blocks with the flow of the model indicated by the arrows as postulated.

Conclusions and recommendations
This study set out to identify the key building blocks that will predict agent performance in contact centres and BPO.Eight building blocks of the human capital predictive model for agent performance were identified from 42 human capital contact centre related articles specific to this study.The building blocks derived from these empirical research articles are: • person-environment fit It is recommended that the interactions of these eight building blocks be tested empirically in a single predictive model in order to understand their role and contribution in a contact centre's overall performance.

Managerial implications
It is suggested that the human capital predictive model serves as an operational management model (see Figure 2 that depicts the flow between the eight building blocks with 22 propositions) that has performance implications for agents and ultimately influences the contact centre's overall business performance.It contributes value in that it explains how P-E fit and JD-R and engagement affects performance; HRM practices (specifically utilising P-E fit for selection purposes and training to develop social competencies) impact agent competence and performance; the various agent competency elements that are to be considered for top performance; and lastly, how well-being and turnover intention impacts agent and business performance.

Limitations and directions for future research
Keywords used to describe the model may have contributed to missing published research.Research was also limited to peer-reviewed business, human resources and industrial and organisational psychology journals found through the EBSCO-host database; additional research may have been found by utilising other databases.
Various propositions, as diagrammatically depicted in Figure 2, can be assessed by future research.The different propositions suggest direct or indirect relationships between variables.It is noted that the first author is currently engaged in a doctoral study to empirically test the human capital predictive model for agent performance in contact centres.

Conclusion
This model building study set out a theoretical (literature) review based on 42 published empirical research articles (for the period 2000-2009) obtained mainly through the EBSCOhost search portal.An eight building-block model emerged that has empirical linkages to agent-and business performance.This model attempts to explain the key phenomena contributing to agent performance in contact centres.Hereby the main objective of the study has been achieved.

FIGURE 2 :
FIGURE 2: A human capital predictive model for agent performance in contact centre.
alluded to the debate about whether routine interactive service work was skilled because it required agents to perform emotion work and articulation work.The amount of task variation, discretion and control was limited and required more clarification.But the work required simultaneous and multifaceted work with people, information and technology; articulation work was often performed within tight timeframes and required the ability to integrate individual tasks into an ongoing line of work (requiring awareness skills, interaction management skills and coordination skills) and collaboration in maintaining the overall workflow.

Table 1
Mouton, J. (2001)fromMouton, J. (2001).How to succeed in your master's and doctoral studies: A South African guide and resource book.Pretoria: Van Schaiks.FIGURE 1:The model building research approach.

TABLE 1 :
Literature search tracking sheet.
P-O fit was operationalised as the correlation between managers' descriptions of the work culture with participants' work preferences.P-O fit added significant incremental variance in predicting employee retention (R 2 = 0.09), but was not related to performance.Limitations included that the extent to which P-O fit predicted retention or performance outcomes was not determined and generalisation to outbound environments was limited as the sample was inbound related.

TABLE 4 :
Human resources management practices.
The increased significance of social competencies within interactive service work gives selection and training greater salience and is used by management to address the indeterminacy of labour, in part, outside the labour process.Limitations include a small sample size (one call centre) with resultant generalisation constraints.

Table 6
continues on the next page → both predicted intention to quit and job complexity and/or variety predicted job satisfaction and affective commitment.Task-related and social stressors were found to predict well-being.The role of emotional dissonance as a stressor in emotional labour was also emphasised.Mukherjee and  Maihotra's (2006)research findings indicated that feedback, participation and team support significantly influenced role clarity which in turn influenced job satisfaction and