The perils of leadership development: unintended consequences for employee withdrawal behaviour and conflict

ABSTRACT

This study examines potential negative human resource consequences of leadership development. Applying theoretical perspectives on job demands and resources, we argue that leadership development may result in potential negative outcomes by introducing new and different demands on workers, requiring adjustment and adaptation. We further argue that the pervasiveness of this risk increases as leadership training is distributed across an organisation. We examine the effects of the implementation of a leadership development programme across a whole organisation (N = 9471), analysing data on incidences of withdrawal (absenteeism) and conflict (formal grievances) over a four-year period. We analyse data using fixed-effects modelling, and our findings indicated that increases in the proportions of leaders undertaking development in nineteen service units of the organisation over this period predicted increases in absenteeism and grievances raised by staff. Implications for theory in leadership development, and the management of development initiatives are discussed.

KEYWORDS:

• Leadership development
• Absenteeism
• Work conflict
• Organisational change and transformation
• Employee well-being

Leadership remains a pressing area of focus for human resource training and development in organisations. Multiple studies have presented evidence for the positive impact of leadership development in organisations, which have been summarised in meta-analyses (e.g. Burke & Day, 1986; Collins & Holton, 2004). Yet, there remain important limitations to our understanding of the impact of leadership development. For example, few studies examine the impact of leadership development on follower or staff outcomes (such as staff well-being, stress and consequent behaviours), rather focusing on the impact for leaders, and few examine objectively measured criteria, rather relying on self-reported criteria. Moreover, the vast majority of studies examine outcomes cross-sectionally or over short time frames (Riggio & Mumford, 2011), leaving a gap in our insight into how the impact of leadership development plays out in the long-term.

These two significant limitations in the literature are especially salient for the study of potential negative consequences of leadership development. Leadership development may be applied in an organisation as a mechanism to enact or cascade culture change for example, involving changes to leadership style across the organisation. Such change invokes processes of adjustment for employees, which are often associated with increased strain or dissatisfaction for some, resulting in higher potential for counterproductive or withdrawal behaviour. If leadership development were to result in such an impact, effects are unlikely to manifest immediately, and may rather take time to emerge.

In the present study, we report an examination of the impact of an organisation-wide leadership development programme (LDP) emphasising performance-oriented leadership styles on objectively measured withdrawal behaviour (i.e. absence), and conflict (i.e. officially raised grievances). The data for our study enable us to observe the impact of leader development on these outcomes across multiple units of the same organisation and over a period of several years. Our study makes three main contributions to the literature on leader development. Firstly, framed around theoretical perspectives on job demands and resources, we are able to observe the effects of leader development on important human resource outcomes longitudinally, analysing trends that emerge as the proportion of leaders completing the training programme increases. Second, we examine objectively collected organisation records, providing much-needed results-level data to the literature on leader development. Thirdly, our data represent a distinctive opportunity to observe the effects in a field setting based on a live leadership programme in a large organisation setting operating in different units. Our results therefore have high ecological validity.

Leadership development and organisational outcomes

Despite the long history of practitioner and organisational interest in development of effective leaders and leadership behaviour, efforts to establish a strong scientific foundation to the field is a relatively more recent effort (Day et al., 2014). The press to better understand the impact of leadership development interventions recognises the need for clearer information about the return that organisations attain for the substantial sums invested in training and other development programmes.

The research literature presents grounds for optimism on the part of organisations applying LDPs. Two meta-analyses notably indicate the positive effects of leadership development. Burke and Day (1986) reviewed data from 1952 to 1982 and concluded that effect sizes were moderately positive. Updating the literature with studies published between 1982 and 2001, Collins and Holton (2004) reported aggregated effect sizes ranging from d = 0.35 to 1.37. Avolio et al. (2009) focused exclusively on experimental and quasi-experimental studies and compared formal training with other forms of leadership intervention. They reported effect sizes of d = 0.60 and 0.67, respectively, further indicating the positive impact of leadership development. Neither of these meta-analyses precisely differentiate criteria in examining outcomes, rather classifying outcomes generally into frameworks (e.g. Avolio et al., 2009: cognitive, behavioural, affective and organisational performance; Collins & Holton: knowledge, behaviour, results; objective vs. subjective). The main focus within the definitions of these categories is individual-level change, for example, in behaviour and attitudes. This is also reflected in the numbers of constituent studies analysed against each outcome category.

These positive findings therefore mask critical gaps in the scope of outcome criteria that have been examined (Clarke, 2012), and in particular in relation to team- and organisation-level effects (Avolio et al., 2009; Collins & Holton, 2004; Riggio, 2008; Sogunro, 1997). This absence of such data diminishes to a degree the confidence that might otherwise be gained from the results of meta-analyses, because effect sizes in those analyses for the most part relate to individual-level change or improvement. It is assumed that those individual-level changes will influence in turn the behaviour of followers, and their performance, contributing to organisation-level outcomes positively.

This assumption may be flawed because change in individual leader behaviour may not result in uniformly positive outcomes at organisational level. Classical contingency theories of leadership (e.g. Path-goal theory; House, 1971) recognise that leadership styles are more or less suited to particular leadership contexts. The effects of behaviour change may produce variable results for different leaders. Furthermore, assumptions of the generalisation of individual-level effects to organisation-level results neglect to consider that leadership change may result in some negative outcomes alongside positive, in particular for workers who may experience stress and strain from adapting to new leadership practices.

Theoretical framing of potential negative effects of leadership development

A key contention in our study is that whilst there is much evidence of the positive effect of leadership development, there are also plausible mechanisms through which potential negative consequences could be observed. Such effects could manifest as undesirable human resource outcomes at team or organisational level as a result of the impact that leaders have on individual staff. In this section, we discuss relevant evidence that leader behaviour may affect well-being generally, and contrast findings of positive effects with critical examination of the contextual factors that may lead to unintended negative outcomes.

To explain the potential mechanisms through which leadership development may result in negative outcomes, we ground our study in theoretical perspectives on job demands and resources (Bakker & Demerouti, 2007). Through applying propositions of the role of job demands and resources on employee stress and well-being, it is possible to explore how leader behaviour, and changes in leadership styles may result in adverse well-being consequences for staff. These effects in turn, may result in greater observation of some of the outcomes of employee stress, such as withdrawal and conflict (Maslach & Leiter, 2016).

There is overwhelming evidence that leader behaviour affects the well-being of individuals they manage (Kelloway & Barling, 2010; Mullen & Kelloway, 2011; Skakon et al., 2010; Kuoppala et al., 2008). Leader behaviour has been shown to predict positive outcomes like employee job satisfaction (e.g. Nielsen et al., 2009), involvement and experienced meaningfulness (Nielsen et al., 2008), happiness and optimism (Bono et al., 2007), well-being (Arnold et al., 2007) and safety behaviour (Mullen & Kelloway, 2009; Zohar, 2002), and ineffective leadership to predict negative outcomes such as emotional exhaustion (Harvey et al., 2007). The proposition that leaders have an impact on their follower’s well-being is therefore well-founded.

Leadership development has been highlighted as a potential intervention for improving occupational psychological health, for the potential benefit of improving well-being-related outcomes such as attendance (Dellve et al., 2007; Kelloway & Barling, 2010). Yet, such propositions arguably neglect the context of leadership development as one of change, in the sense that employees may be faced with changing work environments and experiences as their leaders develop. In this respect, the literature points to the possibility of negative consequences for followers. In a study across different teams, leadership training was found to have mixed results, rather than uniformly positive, on improving employee retention (Wallis & Kennedy, 2013).

In respect of understanding the organisational effects of leader development, the content of programmes is also relevant. For example, workplace health promotion to improve attendance, if focused on individuals as sources of problematic absenteeism (versus e.g. organisational factors) can have the reverse of the intended effect, and actually reduce attendance (Aust et al., 2010; Dellve et al., 2007). To the extent that leadership development emphasised dealing with performance issues solely at the individual level, such development could serve to damage relationships and trust, with consequent negative impacts on well-being.

Such issues are brought into sharper focus when leadership development is employed as a strategy for organisational change. As a component of enabling and manging change (Day, 2000; Higgs & Rowland, 2001) or as a primary mechanism for effecting organisational change (e.g. Boomer & McCormack, 2010; Brown & May, 2012; Turnbull & Edwards, 2005), leadership development is routinely perceived to serve a strategic function (Clarke & Higgs, 2016). Leaders are central to organisational change because they serve to interpret and enact strategic objectives (Schneider et al., 2013), translate them to individual performance objectives (Murphy & DeNisi, 2008) and deal with potential issues of resistance and adjustment to change (Fiol et al., 1999). Moreover, leader actions to shape employee behaviour (Schneider et al., 2013; Schein, 1983) represent a key mechanism through which employees feel the effects of organisational change.

With regard to employee well-being, the experience of organisational change can be difficult for employees (Herscovitch & Meyer, 2002; Rafferty & Griffin, 2006). Framed around models of job demands and resources, experience of organisational change has been proposed to increase worker demands, resulting in greater experience of stress and burnout (Day et al., 2017). From a stress and coping perspective (Lazarus & Folkman, 1984), a principal mechanism that explains the negative effects of change are the requirements for employee adjustment and the associated the psychological uncertainty that it creates, which may be clarified by considering job demands and resources. Firstly, change introduces increased worker demands (Day et al., 2017; Terry & Jimmieson, 2003), including many (e.g. role ambiguity, increased workload) that are associated with stress outcomes such as burnout (Puleo, 2011). Secondly, uncertainty resulting from experiencing change may reduce feelings of control and autonomy (Bordia et al., 2004), which would otherwise act as resources to moderate the impact of increased job demands on employee strain, elevating the risk of associated stress (Terry & Callan, 1997). This may be especially acute for staff compared to managers (Martin et al., 2006), underlining the risk of feelings of loss of control for staff responding to changes in leadership and management styles enacted following leader development.

These mechanisms have the potential to result in negative organisational outcomes. Rafferty and Griffin (2006) found that perceptions of change as being fundamental to the core of the organisation (i.e. transformational) predicted greater uncertainty for individuals, in turn predicting lower satisfaction and higher intentions to turnover. Cynicism may also build as consequence of repeated negative experiences of change over turbulent periods, serving to lower satisfaction and commitment (Reichers et al., 1997). Employee absence is also a significant withdrawal behaviour risk from organisational change (Firns et al., 2006 Kivimäki et al., 2000;). Fugate et al. (2012) examined the role of threat appraisal in absence resulting from organisational change. The extent to which people found change threatening predicted greater incidence of sickness absence.

Alongside withdrawal, conflict may also result from the experiences of changes to management and leadership practices. For example, organisational changes often prompt resistance and increased incidence of grievances (Kirkman et al., 2000). Moreover, changes that result in perceived adjustment to relationships with management (such as may be invoked through leadership development) tend to result in strained relationships between leaders and followers and higher incidence of grievances (Ogbonnaya et al., 2022).

The creation of felt uncertainty and experiences of work environment change accompanying changes in leadership style and behaviour therefore represent potential mechanisms through which leadership development could result in negative outcomes. Changes in leadership behaviour of those undertaking development may serve to change the expectations of employees, resulting in greater uncertainty (Rafferty & Griffin, 2006), and consequently higher stress (Schabracq & Cooper, 1998), ultimately resulting in more negative outcomes for staff and the wider organisation.

Examining outcomes of leadership development: effects over time

Our study is conceptually focused with effects at the macro- or results-level, rather than at the individual level. We accordingly argue that the extent to which the potential negative consequences of leadership development manifest at organisational level is a function simply of the spread or distribution of the programme (i.e. the proportion of leaders in the organisation that undertake development). As more leaders are trained, so the resultant change becomes more widely distributed, creates a greater spread of feelings of uncertainty within the organisation, leading to higher potential for negative effects to become more widespread and frequent.

Observing such an effect necessarily requires examining data longitudinally. Leadership development evaluation research has in the past been typified by short-time frames (Riggio & Mumford, 2011), yet leadership development is a process that occurs continuously over a period of time, rather than episodically in response to a single intervention (e.g. Day & Sin, 2011; Gottfried et al., 2011; Murphy & Johnson, 2011; Riggio, 2008). In the case of organisation-level outcome criteria, we argue that manifest effects on absence and grievance are more likely to occur over a longer time horizon. This is because the impact of leadership development on such criteria result from changes in leader behaviour with followers. In the case of effects on well-being and withdrawal behaviour such as absenteeism, any negative effects of stress and strain are further likely to take time to accumulate as new leader behaviours become more widely distributed in the organisation, before resulting in consequences such as absence from work or counterproductive conflict.

The present study and hypotheses

The context of the present study is relevant to setting hypotheses. The organisational setting for our study is a UK-based local government organisation (UK council). The organisation provides services to citizens within a geographically defined area of the UK (for example providing housing and property services, local taxation, waste and environment management and so forth). In all aspects of this article, the names of relevant units and the organisation are omitted for anonymity.

An organisation-wide LDP was implemented in the organisation over a period of four years (more detail on the specific content of the LDP is included in our method section). The programme was implemented across all units within the organisation, with leaders attending in “cohorts” starting at various points across the period under analysis. The organisation collects objective data on sickness absence and grievances, available for analysis in the present study.

This LDP was undertaken with an aim to develop a more performance-oriented leadership culture. This raises the possibility that individual staff may experience some of the potential negative psychological consequences of adapting to this change in leadership approach and style, especially given that the behaviours that result may tend towards providing challenging feedback to individuals (Cianci et al., 2010).

The cross-time implementation of the programme presents a unique opportunity to examine organisational effects as the distribution of the leadership development increases over time. Following our earlier arguments, we propose that the roll-out of this programme may result in cumulative additional demands on staff, reflecting its performance-oriented focus. Drawing on the job demands-resources model (Bakker & Demerouti, 2007), we expect that changes to leadership style prompted by the LDP will likely introduce greater scrutiny and focus on staff performance, which could affect work demands for staff (e.g. Day et al., 2017). For example, this could include expectations of increased workloads, and through changes to management practices, potential role ambiguity for staff. Furthermore, such changes to leadership practices may increase felt uncertainty (Rafferty & Griffin, 2006), leading to lower perceptions of control at work, accordingly reducing psychological resources for staff. This combination of potential effects on demands and resources would serve as a risk for staff experiencing negative psychological consequences such as stress and burnout (Maslach & Leiter, 2016). We propose that such an effect may be observed in organisational records of sickness absence and grievances in the focal organisation.

We also earlier proposed that where unintended negative consequences may emerge from leadership development, effects would be observed more frequently as more leaders undertake the development, thereby distributing changes to leadership behaviour more widely. In the focal organisation, we are able to examine the effects of the roll-out of the leadership programme on absence and grievance data over time, and our arguments lead us to predict that as more leaders participate in the development, so the incidence of absence and grievances will increase. The organisation structure enabled us to examine these trends within multiple organisational units to examine trends in sickness absence and grievances. We expect that increases in the proportions of leaders in each unit undertaking development (relative to the total number of people in unit) will predict incidence of withdrawal or counterproductive behaviour. We therefore hypothesise:

H1: At unit-level, absence increases as the proportion of managers attending the LDP increases.

H2: At unit-level, grievances increase as the proportion of managers attending the LDP increases.

Method

Data and participants

Data were collected from a local government authority (LGA) in the UK.¹ The LGA was organised into six units, each encompassing several service units responsible for delivering services to the residents under its geographic jurisdiction. The LGA introduced a bespoke leadership programme for all the leaders, focused on developing high performance in the organisation, designed to build greater leader resilience and to enhance ability to empower and manage performance of staff. Programmes for leaders and senior leaders were commissioned. The LGA deemed that the data held regarding their employees were eligible to be used for research purposes, provided anonymity was protected. Data were compiled by officers at the LGA following specification provided by the first author. Data for each individual employee within the LGA was provided including a unique identification code for each employee, their age, gender, ethnicity, nationality, their type of work contract, status as a leader, service-unit membership and for leaders, whether and when they had attended the leadership programme.

Prior to conducting the analysis, data were inspected and cleaned to correct some anomalies. First, the data cleaning revealed a high proportion of employees listed as holding contracts of 0.01 h per week, whom also did not have relevant criterion data for this study (e.g. absenteeism, grievances) and so were not valid for inclusion in the present research. With the support of the LGA, the data were also corrected to ensure that each employee had only one identification code. For example, where employees had duplicate identification codes and they worked half of their time in two different roles or in multiple roles, personnel data were assigned to one of their identification codes or to the role where they work the most hours per week. The second code and its associated data were deleted. Also, there were some discrepancies in the number and titles of services across data sources reflecting the development of the LGA over the focal period, in which service names or structures had changed. These issues were also reconciled providing a group of 19 service units that were identifiable reliably across the data sources.

To map the line management structure of the organisation, information was also gathered on each employee’s line manager, whether their manager themselves had attended the leadership programme, and the team, service and directorate the employee worked in.

After the data cleaning, the remaining sample was 9,471 participants. Of those, 73.6% were female, and the mean age was 44.42 years, SD = 11.65. 54.6% of the sample had a full-time contract and 92.8% had a permanent contract. 74% were British with the remainder representing a wide variety of other nationalities. With regard to the status of the participants, 1,100 were leaders and 263 were senior leaders, who were eligible to attend the leadership programme. Of those, 320 (29.1%) leaders, and 191 (72.6%) senior leaders had attended the programme (513 in total) within the period under study. Two participants had attended both leader and senior leader programmes.

For our analysis on the focal 19 service units, the sample comprised 8,626 participants (75.2% female; mean age 44.39 years). Within the sample, there were 979 leaders, and 196 senior leaders. Of these, 240 (24.5%) leaders and 140 (71.4%) senior leaders attended the programme with the period under study (see also Table 1).

Table 1. Description of the sample at the unit-level (units included in analyses).

Service unit number	Unit	N	N	Absence		Grievances
	Size (N)	JLDP	SLDP	M	SD	M	SD
1.	26	7	3	0.0588	0.1029	0.0007	0.0050
2.	118	16	2	0.2545	0.1536	0.0002	0.0012
3.	37	3	9	0.0983	0.0977	0.0005	0.0036
4.	1032	6	2	0.1869	0.0704	0.0001	0.0004
5.	639	2	5	0.7367	0.1546	0.0005	0.0013
6.	116	11	2	0.5549	0.2826	0.0002	0.0012
7.	107	8	17	0.4016	0.2491	0.0005	0.0022
8.	90	15	6	0.2789	0.2447	0.0002	0.0015
9.	681	6	6	0.6754	0.1975	0.0002	0.0007
10.	812	30	23	0.5233	0.1329	0.0007	0.0012
11.	176	16	7	0.3778	0.1783	0.0005	0.0023
12.	2251	23	12	0.3455	0.1432	0.0002	0.0004
13.	1101	27	13	0.4391	0.1496	0.0006	0.0011
14.	565	37	16	0.2906	0.1300	0.0004	0.0001
15.	302	22	11	0.3004	0.1126	0.0000	0.0000
16.	65	0	0	0.3570	0.1838	0.0006	0.0030
17.	148	4	3	0.2212	0.1454	0.0001	0.0009
18.	102	5	2	0.3355	0.2491	0.0002	0.0013
19.	258	5	1	0.3614	0.1226	0.0001	0.0007

Note: N JLDP = Junior Leadership Development Programme Attendance (cumulative); N SLDP = Senior Leadership Development Programme Attendance (cumulative). Service-level absence was measured as the average number of days absent from work per employee per month. Service-level grievances were measured as the number of grievances raised per employee per month.

The leadership development programme

The LDP was introduced to help establish a more performance-oriented leadership culture in the organisation. The programme content was designed around this objective, and separated across two levels, one for senior leaders (i.e. higher-level service managers) and another for junior leaders (i.e. team leaders and supervisors).

The common features of the programme were a modular format consisting of taught components, periods for reflection, feedback input and coaching sessions. The programme was run in cohorts of approximately 12 participants per programme, with facilitators from an external provider assigned to each cohort.

Both senior- and junior-leader programmes had three phases (see Figure 1). The first phase commenced with a two-day workshop aiming to increase participants’ self-awareness around their leadership impact employing a variety of models including for example, transactional analysis (e.g. Berne, 2016) and resonant leadership (e.g. Boyatzis et al., 2013). Content was selected in particular to help enable leaders to navigate difficult relationships and have more constructive conversations about performance with followers.

At the end of the two-day event, participants were briefed on a 360-degree feedback process, paired with another participant within the cohort, and tasked with gathering 360-degree feedback from nominated people for their partner. After 4–6 weeks, cohorts reconvened for the second phase of the programme, a third input day to share the 360-degree feedback results.

Following this aspect, the content of the leader and senior leader programmes diverged. Senior leaders proceeded to 1–1 coaching sessions, working in pairs to coach each other on work related challenges previously identified in the programme, supported by an experienced coach, from an external provider. In a fourth classroom-based day, senior leaders also considered at a more advanced level ways for dealing with difficult relationships. Junior leaders learned about restorative approaches to conflict and reflect on their role in this process, and also engaged in facilitated peer coaching.

The third and final phase of the LDP was the creation of peer coaching sets. Here participants met in small groups to coach each other on challenges experienced in their work as leaders. The senior leader programme comprised four formal half-day sessions, whereas the junior leader programme comprised three formal half-day sessions. For the purposes of our analyses, the point of attendance on the programme was set at when leaders commenced the programme.

Notably in the design of the leadership programme, there were several aspects that prepared leaders for holding conversations with staff about performance. These included for example, core content on holding constructive conversations about performance and effectiveness and navigating and dealing with difficult working relationships. These aspects highlight potential mechanisms through which changes in leadership style following the programme may, following our proposed mechanisms, result in unintended negative consequences.

Variable specification for analyses

The independent variable is attendance on the LDP and the dependent variables include absence and grievances, which were collected by the LGA on a monthly basis from January 2012 to April 2016. LDP attendance is measured as the proportion of managers participating in the LDP relative to the overall size of the service unit, measured monthly (i.e. N_{leaders attended}/N_{service unit total} in each unit in each month). We assume the effect of the LDP to be cumulative (i.e. as more leaders complete the programme in each service, the proportion increases over time). As the LGA provided information on the type of LDP attended by participants (and given that the content across the two programmes differed), we have created three separate indicators to measure the level of participation in Junior Leadership Development Programme (JLDP), Senior Leadership Development Programme (SLDP) as well as for the overall LDP combining both variants of the programme. Information on absence was provided by the LGA at the individual level. We aggregated the individual-level data to the service level to measure the average number of days absent from work per employee (i.e. unit-member) per month (excluding absence reported by those who have attended the LDP programme). Incidence of grievances was reported at the service level as the number of grievances raised by employees in each month. This was divided by the size of the unit to give a measure of grievances raised per employee per month. A description of absence and grievance data is presented in Table 1.

Analytic techniques

We used fixed effect models to analyse the effect of LDP attendance on absenteeism and grievances. Fixed effect modelling is a widely used statistical technique to analyse longitudinal data where observations are repeatedly collected from multiple individuals or cases, which in our study are the units in the organisation. These models are suitable when there are time-invariant individual characteristics that may correlate with both dependent and independent variables, which can lead to biased estimates (Allison, 2009; Gunasekara et al., 2014). The general form of a fixed effect model can be represented as: $Y_{\mathrm{it}}={\beta }_0+{\beta }_1{X}_{\mathrm{it}}+{\alpha }_i+{\epsilon }_{\mathrm{it}}$Where: Y_it is the dependent variable for unit i at time t. X_it represents the independent variables for unit i at time t. β₀ is the intercept, representing the average value of Y when all independent variables are zero. Β₁ represents the coefficients for the independent variables, indicating the relationship between X and Y. α_i denotes the fixed characteristics of unit i, capturing unobserved heterogeneity. Ε_it is the error term.

The key advantage of fixed effect modelling that makes it appropriate in our study is its ability to control for time-invariant unobserved unit-level factors (i.e. unmeasured characteristics of business units) which may correlate with both independent and dependant variables, and that may otherwise confound or bias analyses (see Longhi & Nandi, 2014). In contrast to traditional multivariate regression analysis or random effect models which do not take account of the omitted variable bias (i.e. variations in the dependent variable are caused by unmeasured characteristics of the unit which correlate with the independent variable), fixed effect models focus on within-unit change over time, thereby removing the confounding influence of all time-invariant unit-level factors such as geographic location, function area, initial baseline measurement, and so on. Moreover, the nature of our IV (programme attendance) as an externally-driven “event” which was not consequent to our DVs (absence or grievance rate), and which occurred over varying time intervals across units (because the programme was not implemented over equal time intervals across units) makes our specification of the regression especially suitable compared to for example, examination of standard analyses of crossed-lagged effects.

To further support our analytic approach, recognising that both fixed and random-effects modelling have advantages and drawbacks (Longhi & Nandi, 2014), the appropriateness of each approach for specific data sources can be determined through the Hausman test (Hausman, 1978). Results of the Hausman test for our data confirmed that fixed-effects models were most appropriate (details of these analyses available on request from the first author). The specification for our fixed-effects analysis were based on models in STATA using the “xtreg” command with the “fe” option (Longhi & Nandi, 2014). Accordingly, the STATA function created unit-specific dummy variables and included them in the regression model to control for their fixed-effects. Since the analysis is carried out at the service unit level, it is not affected by the constraints of hierarchically structured data (e.g. individuals nested within the same unit tend to share more similarities with one another compared to those who work in different units).

In summary, and to ensure clarity in the interpretation of our results, the fixed effect models we report focus on how change in the dependent variable over time is affected by change in the independent variable over time within the same service unit. In short, our analyses test more directly whether given its fixed characteristics, a service unit experienced more adverse effects of absence and grievances over time as a greater proportion of leaders completed the LDP.

Results

We started by examining the overall effect of LDP attendance on absence and grievances and then repeated the analysis for different types of LDP to test whether they have different effects on employee outcomes. The results of fixed effect regressions for the overall LDP attendance are shown in Table 2. The second column of Table 2 shows that LDP attendance has significant and positive effects on absenteeism (b = 0.90, p < 0.001), which suggests that as the proportion of managers attending the LDP increases over time, staff absence correspondingly increases. There is also a significant effect on grievances (b = 0.01, p < 0.001), indicating that the number of grievances lodged by employees tend to rise as more managers attend the programme.

Table 2. Fixed effect regressions of LDP on absence and grievances.

	Absence	Grievances
Attendance	0.90***	0.01***
	(0.00)	(0.00)
Constant	0.38***	0.00
	(0.00)	(0.45)
R^2	0.06	0.03
N (unit-month observations^a)	570	570

Note: p-values in parentheses.

^a N represents unit-month observations (i.e. not leader sample sizes).

* p < 0.05, ** p < 0.01, *** p < 0.001.

Next, we tested the effect of JLDP and SLDP, separately. The results presented in Table 3 show that absence is significantly associated with attendance of both JLDP (b = 1.36, p < 0.001) and SLDP (b = 1.17, p < 0.001). Similarly, attendance at the JLDP (b = 0.02, p < 0.01) and SLDP (b = 0.01, p < 0.01) have significant effects on grievances.

Table 3. Fixed effect regressions of JLDP and SLDP on absence and grievances.

	JLDP		SLDP
	Absence	Grievances	Absence	Grievances
Attendance	1.36***	0.02***	1.17***	0.01**
	(0.00)	(0.00)	(0.00)	(0.01)
Constant	0.39***	0.00	0.39***	0.00*
	(0.00)	(0.10)	(0.00)	(0.04)
R^2	0.05	0.01	0.03	0.01
N (unit-month observations^a)	570	570	570	570

Note. p-values in parentheses

^a N represents unit-month observations (i.e. not leader sample sizes).

* p < 0.05, ** p < 0.01, *** p < 0.001.

Figure 2 illustrates the trends in absenteeism over time for service units with high, medium and low levels of participation in the LDP programme.² It can be seen that absenteeism has generally increased over the period during which the LDP programme was rolled out. The service units where fewer leaders had attended the programme reported higher levels of absenteeism. These results could have led to the erroneous conclusion that participation in the LDP programme improves employee outcomes if we had estimated the effect of LDP attendance based on traditional linear regression methods. However, our focus on within-unit pattern of change over time has revealed a rather different story. As the fixed effect models in Tables 2 and 3 have shown, staff absence actually increases as the proportion of managers attending the LDP increases within the unit. These results highlight the importance of controlling for the presence of unobserved heterogeneity which can threaten the validity of causal inferences in longitudinal research. Our analyses provide full support for Hypothesis 1 and Hypothesis 2, as attendance in both JLDP and SLDP predicts increase in absenteeism and increases in grievances.

Figure 1. Overview of the LDPs.

Figure 2. Trends in absence among service units over time with high, medium and low level of participation in the LDP. Note: x axis represents months. y axis represents absence per employee per month.

Discussion

The continued popularity of LDPs in organisations, and the benefits that are sought by those that commission them, create a pressing need to fully understand their potential consequences. While meta-analyses of the outcomes of leadership training demonstrate the potential benefits of such programmes, we have proposed that leadership development may result in unintended consequences, some of which may be negative. Drawing on mechanisms of theories of job demands and resources (Bakker & Demerouti, 2007), we argue that negative outcomes may stem from adjustment of employees to increased demands of adjustment to changes in management practices and lower control characterising uncertainty brought about by new leadership styles. We further propose that the potential for negative effects become more prevalent as the distribution of the development increases. Our study of the impact of leader development on absence and grievances across a whole organisation over a period of 4 years, demonstrated the potential for such negative consequences. At the unit-level, we observed increased absence and grievances as more leaders completed the development programme. We discuss these findings, considering theoretical, future research, and management implications.

Potential negative human resource outcomes from leadership development: theoretical and conceptual implications

There is limited theoretical explication of the macro outcomes of LDPs, with multiple calls for more examination of organisation-level outcomes (Avolio et al., 2009; Collins & Holton, 2004; Riggio, 2008; Sogunro, 1997). In our paper, we extend this criticism to argue that leadership development research has to date failed to adequately consider or explain the potential for leadership development to result in adverse human resource outcomes.

To address this gap, we proposed that leadership development creates changes in management style that could result in negative outcomes. Framed in theoretical models of job demands and resources, the explanatory mechanism reflects the process of adjustment, which has been related to experience of stress and strain (Rafferty & Griffin, 2006). The LDP that we examined in the present study was designed to change leadership culture to become more performance-oriented in the organisation. The underlying objective of the programme was to create better service provision through improved performance management and confronting poor performance more directly.

A potential mechanism from this kind of development programme to negative consequences such as absenteeism and conflict is therefore through the creation of greater job demands (e.g. higher workload, role ambiguity and management intervention), alongside lower autonomy and control resources resulting from uncertainty, leading to potential strain on employees who need to adjust accordingly. While our data did not include direct measures of employee attitudes and well-being, a logical consequence of the proposed mechanisms is increasing levels of negative outcomes as more leaders across the organisation completed the training. At the macro level, the potential for negative consequences resulting from employee adjustment should increase as more leaders undertake development, leading to greater incidences of absence and conflict. Our cross-organisational data collected over a 4-year period enabled us to test and confirm this effect for employee absence and grievances.

Our results are important to caveat in three ways. Firstly, we examined a single programme as it was deployed over several years, and we have noted that the focus of this programme was to develop a more performance-oriented culture. Therefore, compared to alternatives, this programme may have especially salient features that could provoke some of the mechanisms we highlighted in our theoretical arguments, leading to the outcomes we observed in the data. In short, it would be important to replicate our findings by examining other leadership programmes. Nevertheless, the programme design shared many of the features of typical leadership development provision (Day, 2000) and was not restricted to human resource management practices such as performance management and review. Secondly, although we are able to model the effects of increasing distribution of the programme in the organisation, it is important to acknowledge that we could not measure the extent to which learning from the programme was transferred to work practices by leaders. Training transfer is a critical issue in learning and development more widely (Burke & Hutchins, 2007). Thirdly, we recognise that the use of single-organisation design means that the possibility of a third unmeasured organisational contextual factor, or external environmental factor (e.g. worsening economic climate) affecting participation in the leadership programme and absence and grievances, cannot be ruled out through our analyses. This is plausible given the nature of organisational development generally, and the deployment of interventions in response to external demands. However, although solely based on our understanding of the focal organisation in this study, we are not aware of such factors and to our knowledge, the organisation was not subject to other sustained organisational changes during this period apart from the LDP. Taking account of these caveats, our analyses nevertheless did demonstrate adverse effects associated with greater distribution of the leadership programme.

Our findings therefore do highlight some of the potential macro outcomes of leadership development that might be unanticipated. Researchers and practitioners more commonly evaluate leadership development through the lens of positive outcomes such as improved performance, satisfaction or relationships. Our results suggest that leader development could also be accompanied by undesirable consequences, and offer potential theoretical explanations for their emergence (i.e. increases in job demands and reduction of control for staff resulting from changes to management and leadership behaviour, accompanied by wider distribution of the programme).

Balancing positive and negative outcomes of leadership development

Incidence of absenteeism or grievances is on a superficial level, undesirable. Managers are obviously unlikely to set out on any kind of organisation development programme with the intention of raising these kinds of counterproductive outcomes. However, on the other hand, problems of adjustment do not diminish the strategic need for change.

This raises the prospect that managers (especially top management teams initiating strategic change) may take an ambivalent view of negative outcomes such as absenteeism and employee grievances. For example, when moving to a more performance-oriented leadership culture, it may be necessary to introduce greater scrutiny and more detailed measurement of performance and productivity. This could result in performance feedback for some staff that challenges underperformance that may have in the past gone unchecked. In a previously feedback-light environment, hearing the message of the need to improve may be challenging for some (Kinicki et al., 2004).

In our study, we examined two unit-level manifestations of the negative consequences from staff experiencing such situations. Firstly, increased levels of strain and employee stress were proposed to result in higher incidence of absenteeism over time (Kivimäki et al., 2000). Secondly, changes to perceived relationships between leaders and staff may erode mutual trust and lead to deteriorating employment relations, change or reduce perceived social resources in the workplace, thereby raising the prospect of employee grievances (Ogbonnaya et al., 2022). Such outcomes are further problems to deal with, but from another perspective could be seen as a necessary consequence of leadership development. Positive outcomes may thusly be balanced or traded-off with negative outcomes of leadership development.

Three issues are highlighted in respect of this observation. Firstly, there is need and space for theoretically-informed research into the effects of leadership development. Our study highlights avenues for future study through the lens of job demands and resources. For example, our study has enabled longitudinal observation of the association of adverse effects manifesting within organisational units as leadership development was rolled out in the organisation. However, future studies could apply the theoretical perspectives more explicitly by capturing individual-level data on perceptions of job demands, uncertainty, stress and relations with leaders and management. This would enable insight into the multi-level psychological effects of leadership development, and potentially understanding of the differentiation of positive and negative outcomes of development programmes.

Secondly, the issue of time is brought back as a critical consideration for the impact of leadership development. Leadership theory has been criticised for neglecting to adequately incorporate time in conceptual development (Shamir, 2011). In the case of leadership development undertaken to promote or support organisational transformation, the true extent of effects may take many years to realise. For example, our data show rising absence as the programme implementation was more widely distributed. However, as programmes continue to be implemented and culture change stabilises, a levelling, or return to baseline for absence and grievances might be observed. Moreover, in time, improvements in the ways in which absence and grievance are managed may also be observed. Put simply, after changes have “settled-in,” and people have adjusted, positive benefits may be seen more clearly. This further lengthens the timeline horizon for studying outcomes of leadership development. In short, our findings underline the need for long-term evaluation of leadership development beyond short-term behaviour change in leaders themselves, and their followers.

However, in contrast, thirdly, given that our data did not permit identification of any specific differentiation of effects across employees based on their performance, the prospect remains that increased absenteeism and conflict is a more generalised problematic outcome of leadership development. That is, there is a risk that adjustment, coupled with potentially draining effects of experiencing more demands caused by absence of colleagues, and a high-conflict environment could create a snowballing effect of rising adverse behaviour and outcomes. This is a considerable potential risk for organisations, raising the question of how to mitigate the risk in the design of LDPs.

Designing leadership programmes to reduce potential negative outcomes

There are three direct steps that LDP designers and facilitators could take to address the potential risk of adverse outcomes. The first key step is to acknowledge and invite leaders to analyse the ways in which negative outcomes might manifest. A possible focus for this analysis is to consider the extent and nature of follower adjustment that might be required as leaders apply new learning in work with their teams. Leaders could give individualised consideration to how specific team members (assuming their span of control permits) might respond to new practices, feedback about performance, or changes to work demands. Team-member relationships could also be considered to anticipate risks that colleagues’ motivation might be affected by the reactions of others around them. Strategies might be developed to transfer learning to leadership activities in a way that effectively anticipates and manages reactions. This could help to reduce worker uncertainty and increase feelings of autonomy.

A second step is to incorporate material on employee uncertainty, adjustment, stress and well-being more directly into leadership development. Commentary in the literature on the most common content of leadership development (Day, 2000; Day et al., 2014) notably omits to highlight the inclusion of content on the management of staff well-being and stress and understanding of the consequences. This would seem to be a significant oversight in the ways that leaders are prepared for their roles. If leaders were more informed about the signs and manifestations of poor well-being and effective strategies for dealing with them, then any arising negative consequences of leadership development could be identified early and addressed. There are wider implications of this issue supporting a greater role for disciplines focusing on well-being effects of work (e.g. occupational health psychology) in the development and preparation of leaders.

Thirdly, with respect to conflict and grievances, human resource professionals overseeing leadership development could also put in place structural solutions to support managers and leaders and manage communication and participation with staff. This could, for example, include establishing better involvement of staff in developing strategies for leadership development or programme content, helping to improve and maintain trust and employment relations. Peer or senior managerial support mechanisms could also be put in place to help deal rapidly with issues of staff conflict before they escalate. This would be especially critical if the planned distribution of the development programme covers a high proportion of the organisation.

Study limitations

There are some limitations of our study to note in interpreting the implications of our findings. Firstly, we recognise that our results are based on data from a single organisation, and as discussed earlier, based on a single LDP. This limitation frames the caveats that we highlighted around our findings, specifically around the nature of the leadership programme and the potential effects of unmeasured organisational factors. A multi-organisational survey could capture data on the extent to which negative outcomes such as absenteeism and grievance result from leadership development generalise across different organisational settings, and moreover could examine moderating effects of programme content. However, such methodology would have significant limitations compared to our access to and analyses of objective data on programme attendance, organisation structure and make-up and HR records. This puts in perspective a strength of our design in that we test the impact of leader development across a whole organisation in a field setting.

A second notable limitation of our study is the nature of the data that we analyse. We are able to analyse extensive records within the organisation concerning absence, management structure and attendance at the leadership development, enabling a macro-level analysis of the impact of increasing distribution of the programme. However, as earlier highlighted, there are individual-level variables that would strengthen our testing, for example felt uncertainty, psychological well-being, stress and strain would be informative to understand the effects in more detail. Moreover, to address wider questions raised in our discussion about the trade-off between performance gains and negative consequences, data on performance are needed. Nevertheless, the impact of leadership training on performance is evidenced in past studies. By contrast, our study is novel is proposing possible ways that leadership development may result in negative outcomes, thereby setting a foundation for more detailed multi-dimensional and multi-level modelling of leadership development outcomes in the future.

Conclusion

Leadership development seems to be a ubiquitous feature of organisational life, especially in medium-size and bigger enterprises. Indeed, many readers of and contributors to the literature on leadership development will likely have their own experiences of such programmes. The intended purpose of leadership development is invariably positive, whether to enhance organisational competence and capability in leadership, improve motivation and engagement or develop performance. However, the data reported in our study also highlight the risk of a potential downside; that leadership development may have unintended consequences, which in the organisation we studied comprised increases in employee absence and grievances as more leaders undertook the programme. The unintended negative HR consequences constitute some perils of leadership development. It may not be possible to fully avoid them in the course of implementing leadership programmes, but our study provides some foundation direction for ensuring we understand them fully and address them more effectively in the design and management of leadership development.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Given the nature of the variables we examine, it is relevant to point out that data for this study were collected some years prior to the COVID pandemic.

2 Note that we do not present a figure from the data on grievances; given the episodic nature of grievances, graphical representation is less illustrative than absence data.