Human Resource Differentiation and Labor Market Segmentation: A Systematic Review and Research Agenda

Table 1 lists the publications included in the review. Most of the analysed articles presented quantitative results, and the majority used linked employer-employee data. A dual division of labor market and organizational staff was usually evident. This approach was reflected in the theories and models used (Core-periphery model, Dual labor market theory, Insider-outsider theory). Studies on Western Europe predominated.

In most cases, HR differentiation was analysed as an effect of LMS. Some authors (e.g., Kim, Kwon, & Kwon, 2020) noted that labor market segments influence company-level decisions regarding particular employee groups. Four articles focused on individual employer decisions to differentiate employees, which aggregated into labor market segments, and five articles highlighted the mutual influence of HR differentiation and labor market segmentation. The latter articles (e.g., Baron & Bielby, 1980) noted that institutions and segmentation in the labor market influence company decisions, thereby altering or perpetuating the labor market division.

Our analysis led to some interesting conclusions regarding the properties of HR differentiation and the interplay between HR differentiation and LMS. In the following presentation of the results, we will first demonstrate our findings regarding the properties of HR differentiation and then present a model describing the dynamic relationship between HR differentiation and LMS.

The review yielded three findings relevant to understanding the interdependency of HR differentiation properties with LMS, which will be presented below. Details for this part of the analysis can be found in Table S5 in the Supplementary material.

First, some properties were understood differently in the articles reviewed than in the model of Marescaux et al. (2021). For example, the basis of HR differentiation included not only individual job-based (e.g., permanent/temporary contract; Pfeifer, 2009) and person-based (e.g., ethnicity; Friberg, 2012) criteria, but also criteria relating to differences between branches in terms of costs generated (Pulignano et al., 2020) and differences among departments, particularly their different rhythms of work as these affect the needs of staff in terms of working time organization (Ilsøe, 2012). Additionally, formalization was often understood in a different way than in Marescaux et al.’s (2021) work, in which it was described as the degree to which HR differentiation was grounded in formal management-led policies and practices. In the articles reviewed, formalization tended to describe the degree of rigidity in organizational resource allocation decisions resulting from, among other things, labor law or collective agreements (e.g., Ilsøe, 2012; Ochsenfeld, 2018). This way of understanding formalization highlights the role of various actors (e.g., trade unions) in influencing the unequal allocation of resources within an organization.

Second, during the analysis, two new properties were identified: distribution shape, which refers to the distribution of organizational resources allocated to employees (e.g., Suleman et al., 2019; Martin et al., 2018), and central tendency, which depicts the central value of resource distribution that makes it possible to specify the position of an employee or a group of employees in a larger collective (Suleman et al., 2019). Distribution shape was typically operationalised as the ratio of core to peripheral employees (e.g., Polavieja, 2003) or as the skewness of total wage (Suleman et al., 2019). This property is particularly important in explaining employee attitudes since employees may perceive their status in the organization differently, depending on the number and size of employee groups. For example, permanent staff might change their behaviour in response to a shift in the number of employees on temporary contracts (Ochsenfeld, 2018). The average or median value often measures central tendency. It should help to understand employee attitudes when considered alongside other properties such as outcome heterogeneity (i.e., the magnitude of the differences in a given group of employees). For example, along with extreme values, employees might consider a central value in the resource distribution within a given group to determine those with whom they will compare themselves (Putnam-Farr & Morewedge, 2021). In addition, a higher average level of resources (e.g., pay) allocated to employees may attenuate the negative consequences of differentiation (Schmidt, Pohler, & Willness, 2018).

Third, the review revealed interesting interrelationships between properties of HR differentiation. For example, Suleman et al. (2019) noted that the central tendency level is related to the particular distribution shape. Heinze and Wolf (2010) demonstrated that central tendency is linked to outcome heterogeneity. Finally, Ochsenfeld (2018) mentioned that a change in distribution shape affects outcome heterogeneity. Considering that these three properties (central tendency, distribution shape, outcome heterogeneity) reflect three characteristics of statistical distribution – central tendency, shape, and dispersion (Trochim, 2022) – one can conclude they form a group. This group of properties is distinct from that identified by Marescaux et al. (2021). The latter group, consisting of basis, formalization, and purpose, addresses the process of allocating resources. Consistent with the terminology of Marescaux et al. (2021) and Martin et al. (2018), we label the first property group as the outcomes of HR differentiation. In contrast, we describe the second property group as the process of HR differentiation. Note that we do not assign the property resources to either of the two groups, as each of the other properties can be considered in the context of some or all HRM practices. Therefore, resources can be classified in both the former and the latter group.

Based on our literature review, we developed a model to describe the relationships between HR differentiation and LMS. This model, shown in Figure 1, can serve as a starting point for explaining their complex interactions and planning further research.

Figure 1.

Model of the relationships between human resource differentiation and labor market segmentation.

*The resources property can be classified into both groups as it is considered in the context of each of the other HR differentiation properties – they can refer to one (or all) of the three groups of organizational resources described in AMO model (resources that enhance employee abilities, motivation and opportunities; see Bos-Nehles et al., 2023).

In this model, external determinants shape the plans of management and HR departments for differentiated resource allocation (i.e., intended HR differentiation) [A1] (e.g., Chen & Tang, 2022; Hakim, 1990; Kalleberg, 2003). They also determine the local, regional, and national labor market situation, thus influencing LMS. From the individual company perspective, LMS appears as a current labor market situation that determines the quantity and quality of human capital available to the company as well as employees’ workplace expectations and bargaining position [A2] (e.g., Chen & Tang, 2022). Interestingly, the direct effect of a given external factor on HR differentiation may have different effects than an indirect influence through the LMS. One example is the impact of legal conditions on working arrangements in organizations. In general, collective bargaining agreements increase the formalization of HR differentiation, leaving less freedom for employers to differentiate employees (Heinze & Wolf, 2010). Consequently, multi-employer collective bargaining systems (e.g., at the sector level) should reduce the outcome heterogeneity (e.g., van Jaarsveld et al., 2009). However, strong legal protection for some groups of employees (e.g., those on standard contracts) may enhance LMS and encourage employers to increase the number of people employed under inferior conditions (i.e., people recruited from the peripheral segment of the labor market) as companies seek both numerical and functional flexibility (Pulignano et al., 2020). This approach may change the distribution shape and ultimately increase outcome heterogeneity. Furthermore, it should be added that the impact of external factors on the differential allocation of organizational resources is often linked to the specific legal and social conditions of individual countries. For example, Pulignano et al. (2020) point out that the differences in inequality found in subsidiaries of the same multinational company operating in different countries are due to the labor legislation and industrial relations systems.

In our model, we present not only the impact of external but also internal factors on the intended HR differentiation [B1]. For example, in high-tech organizations, employers mainly focus on retaining employees in whom they have invested heavily through training. In this case, the purpose of HR differentiation is to increase differences between more and less skilled employees, increasing expected outcome heterogeneity (Avent-Holt & Tomaskovic-Devey, 2014; Suleman et al., 2013). These internal factors interact with external factors, but do not dictate the decisions of managers, who always have a degree of discretion in choosing a particular personnel management strategy given the particular circumstances (Pulignano et al., 2020).

Intended practices become actual HR differentiation when middle and lower-level managers diversify the resources allocated to their subordinates. The transition from intended to actual HR differentiation, however, is not a straightforward process (Marescaux et al., 2021) as many external [A3] and internal [B2] factors exist, leading to inconsistencies between planned and implemented HRM practices (Makhecha et al., 2018). For example, labor law harmonising some of the working conditions for all employees, similar to collective agreements, should lead to a greater formalization of HR differentiation (e.g., Chen & Tang, 2022). However, labor law regulations are not always enforced, and internal factors may play a greater role; thus, the intended formalization may differ from the actual one (Chen & Tang, 2022). Those internal factors may consist of, among other things, social demographics such as gender, race, or organizational position used by managers. Managers using such categories may allocate resources in a different way than their supervisors or HR professionals have assumed (Avent-Holt & Tomaskovic-Devey, 2012).

Actual HR differentiation influences employees’ perceptions about unequal resource allocation; however, this process is strongly affected by the company’s internal factors [B3]. One is the size of the organization. Larger companies generally pay more (Cobb & Lin, 2017; Suleman et al., 2013); hence, the central tendency in pay within these organizations (although not necessarily in each department/team) is usually higher than in smaller entities. Moreover, large organizations are characterised by a greater reduction in wage inequality (captured by outcome heterogeneity) (see Cobb & Lin, 2017; Heinze & Wolf, 2010; an exception to this rule is described by Kim et al., 2020). It is therefore possible that in these companies, employees are more likely to accept HR differentiation (irrespective of the actual allocation of resources).

In our model, at each level of HR differentiation (intended, actual, perceived) we highlight the interdependencies that exist between two groups of properties (process and outcomes of HR differentiation), but do not assume a specific pattern of relationships between them ([C1], [C2], [C3]). In the case of actual HR differentiation, we found publications suggesting that the differentiation process influences outcomes (e.g., van Jaarsveld et al., 2009); however, it is difficult, based on the reviewed articles, to describe unambiguously the complex relationships between these two groups. For this reason, we do not discuss the interaction of the two groups further here and move on to present the most important element from the perspective of the relationship between HR differentiation and the LMS, namely, the dynamics of HR differentiation.

The dynamics of HR differentiation concern both the actual allocation of resources [D1] and the attitudes of employees resulting from their subjective perception of HR differentiation [D2]. Three aspects of this dynamics help understand its effects. First is the stability of the allocation of resources over time. It determines how the long-term distribution shape evolves and whether one can even claim intra-organizational segmentation. If subsequent HR differentiation decisions consistently move towards the distinguishing of several groups of employees with different amounts and/or rules for resource allocation (as, for example, in exclusive talent management; Kwon & Jang, 2022), we should expect employment segments to emerge. However, not every HR differentiation decision leads to a distinction between employment segments; sometimes one group of employees receives more resources in a particular area of HRM and a different group in another, so the differences between them level out (Piasecki, 2020).

The second issue pertaining to the dynamics of HR differentiation concerns the internal boundaries between emerging employment segments. Only when employees find it difficult to achieve a significant alteration in the amount of resources obtained in relation to other members of the organization (i.e., a change in the intra-organizational segment) can we speak of a certain persistence of differences in resource allocation. In these situations, certain groups of employees (e.g., women) can be trapped in inferior employment segments (e.g., Kalleberg, 2003; Suleman et al., 2019), lacking the characteristics or resources to move to better segments. However, these internal boundaries can be reinforced by employees themselves. For example, representatives of employees known as ‘insiders’ (e.g., those with a permanent contract), being aware of the privileged position of those they represent, may negotiate changes whose costs will be borne by ‘outsiders’ (Ochsenfeld, 2018).

The third element of the dynamics of HR differentiation is the mechanism of self-fulfilling prophecy. This phenomenon can result from the attitudes of managers and employees. Managers tend to allocate more resources to employees who are perceived as more able (or, for other reasons, are classified as primary labor market employees), which gives these employees more opportunities to strengthen their position in the organization and eventually achieve above-average results (van Dijk et al., 2020; van Ophem, 1987). At the same time, employees with an advantageous position tend to feel safe and may plan long-term investments in, for example, their human capital. By contrast, those in disadvantageous positions often try to accustom themselves to an insecure situation and take a short-term perspective on their future (Loveridge & Mok, 1979). Eventually, this leads to increased differences between employees belonging to superior and inferior segments.

The transformation process of intra-organizational segments into labor market segmentation happens in at least two ways [E]. In the first scenario, companies behave similarly as a result of imitating each other and being subjected to the same conditions. As a result, in a given market, organizations have a similar, relatively stable distribution of employees with different working conditions such that labor market segmentation goes across company boundaries (Hakim, 1990; Kalleberg, 2003). In the second scenario, employees classified as peripheral in companies offering good working conditions, while receiving little support from the organization, perform (in line with the self-fulfilling prophecy mechanism discussed earlier) increasingly worse in relation to core employees. Over time, this results in their reallocation to poorer jobs (due to their own decisions or those of their superiors) and makes organizations offering better working conditions dominated by core employees (van Dijk et al., 2020). In other words, the boundaries between labor market segments begin to run along company boundaries. Of course, it is also possible that organizational segments will not develop into labor market segments. They will not, for example, when organizations use different bases or purposes (i.e., criteria or goals) for differentiated resource allocation. Given the previous analysis, such a situation is most likely to occur when external factors that conspire to make organizations similar have little impact.

First, the review helped to structure the field of HR differentiation research. The analysis revealed two new properties (distribution shape and central tendency) and proposed a division into two groups of properties (outcomes and process of HR differentiation). These new constructs should be the starting point for further research into the properties of HR differentiation, a need that various researchers have recently highlighted (e.g., Piasecki, 2020; Rofcanin et al., 2019).

Second, this study provided a better understanding of the impact of external and internal factors on (intended, actual, and perceived) HR differentiation. As mentioned in the introduction, there is a growing need to employ more contextual sensitivity in HRM research, and this also applies to HR differentiation research (see e.g., Luo et al., 2021), especially considering its macro consequences. This research represents a step forward in better using the economic developments in HRM (see Kaufman, 2020).

Third, the review guided research into HR differentiation dynamics. By detailing its three components (stability of employment segments, internal boundaries, and self-fulfilling prophecy), our analysis provides a better understanding of the process by which employment segments emerge from HR differentiation. The analysis is particularly timely as scholars from different research strands have been indicating the need to analyze changes in HR differentiation to better understand its consequences (e.g., Kwon & Jang, 2022; Piasecki, 2020). This is especially relevant in the context of the various phenomena that are dramatically changing the labor market and individual organizations today, such as the rise of the gig workforce, the use of AI in HRM, or the pressure from society to address sustainability in employee matters (Brodzik et al., 2023). An example of how our model can be used to analyze such phenomena is presented in the ‘Practical implications’ section.

Finally, this is the first study to capture the relationship between HR differentiation and LMS comprehensively. The model created based on the review allows us to better apprehend the processes of mutual reinforcement between the intra-organizational and market segments of employment and also the relationships that can undermine these processes (such as attitudes of employees or their supervisors). This is particularly important given the need for an in-depth understanding of inequalities within organizations and at the labor market level (Cobb & Lin, 2017; Peters & Melzer, 2022).

A first direction for future research should be to explore more broadly the impact of newly identified HR differentiation properties (distribution shape and central tendency) on employee attitudes and company performance. As some authors have highlighted (e.g., Ochsenfeld, 2018), this exploration should cover a broad catalogue of organizational resources that include not only motivation-enhancing resources, in particular pay and the related financial elements, which dominated in our analysis (see Table S5 in the Supplementary material).

Second, it would be worth investigating how properties describing the two groups distinguished in our analysis affect each other (paths [C1, C2, C3] in Figure 1) on each level of analysis. Our review revealed that some authors consider the relationship between outcomes and process to be causal, such that the former is an effect of the latter. However, an inverse relationship may occur at the level of employee perception. Employees may first analyze the structure of differences and then attempt to interpret them in the context of management’s purpose and differentiation basis (see Sumelius, Smale, & Yamao, 2020, for an example of the distinction between immediate reactions and sensemaking about talent status).

Third, research should look at the simultaneous impact of internal and external (especially LMS) antecedents of HR differentiation (paths [A1, A2, A3, B1, B2, B3] in Figure 1). In-depth analyses would enable us to understand how boards and HR departments plan for differentiated resource allocation i n response to the influence of both groups of factors and how much attention they pay to LMS. Moreover, in a time of growing interest in the inconsistencies between intended, actual and perceived HRM practices (Makhecha et al., 2018), it would be worthwhile investigating how internal and external factors interact with the process of implementing and interpreting differentiated resource allocation (bearing in mind that their impact may vary according to the group of HR differentiation properties or even individual properties). This research would be best carried out in other parts of the world than Western Europe, which dominated the current review. Note that the historical, social, and economic conditions of countries from other regions (e.g., from Central and Eastern Europe) can strongly influence not only the factors affecting HR differentiation (such as labor law), but also the implementation and employee perceptions of differences in resource allocation. Research in multinational corporations could verify whether employees’ attitudes towards HR differentiation are due to their background or rather to the applied HRM practices. For example, Friberg (2012) describes a discrepancy between employees and employers in interpreting HR differentiation occurring in Norwegian companies employing Polish migrants. While Norwegian employers attributed HR differentiation to differences in work culture, Polish migrants perceived it as a result of the different bargaining power of individual employee groups.

Fourth, the time factor cannot be overlooked when investigating the link between HR differentiation and LMS (paths [D1, D2] in Figure 1). For example, examining self-fulfilling prophecy effects and the trapping of certain groups of employees in inferior segments (e.g., Kalleberg, 2003; Suleman et al., 2019) constitutes an essential line of inquiry. A more thorough understanding of how HR differentiation changes over time and sometimes develops into permanent workforce segmentation is important not only from an academic point of view but a practical one (see below). Using mixed methods (quite rare in the articles reviewed) may prove helpful.

Finally, a society-relevant line of research would be to analyze more deeply the influence of HR differentiation on LMS (path [E] in Figure 1). This need was also articulated by some authors in the review (e.g., Suleman et al., 2019). One way to achieve this goal would be to analyze data from companies in the sector where LMS has already been studied and obtain data on actual or perceived HR differentiation. The theories and models highlighted in this review, such as relational inequality theory (e.g., Tomaskovic-Devey et al., 2015), can provide a theoretical basis for this type of analysis. Other constructs related to LMS and HR differentiation are also worth considering, such as precarious work (e.g., Allan & Blustein, 2022). It would be interesting to examine the conditions under which differences within an organization and on a specific (e.g., industry) labor market lead to the formation of precarious employment. Note that, since both HR differentiation (Lepak & Snell, 1999) and LMS are dynamic phenomena (Eichhorst & Kendzia, 2016), research over a longer time horizon seems necessary for these issues.

This literature review should be of interest to practitioners for at least three reasons. First, by providing an extended catalogue of HR differentiation properties, it allows for better planning of organizational resource allocations. It can also provide the basis for developing more sophisticated indicators measuring equality in the workplace, which are now expected of organizations (Brodzik et al., 2023). In this way, management should be able to more easily communicate to employees the reasons for differential resource allocation, emphasizing the concern for overall fairness in the workplace. Moreover, an analysis of changes in these indicators over time would make it possible to identify more quickly the ‘entrapment’ of certain groups of employees in inferior positions (e.g., Kalleberg, 2003; Suleman et al., 2019) and consider appropriate remedial action (e.g., retraining, relocation, change of remuneration system). This should help avoid the self-fulfilling prophecy problem.

Second, our model provides a better understanding of changes in the differential allocation of organizational resources under the influence of (dynamic) external and internal factors and their effects. This is particularly vital for organizations operating in various markets, experiencing different legal conditions or those with blurred boundaries, such as platform organizations (Luo et al., 2021). For example, the model could be used to track the impact of more intensive use of AI at work (Brodzik et al., 2023) by first analysing how it will affect management decisions regarding resource allocation (e.g., whether a strategy of greater support for employees using AI will be adopted), then how it will affect managers’ practices (taking into account, for example, their technological concerns) and employee perceptions (e.g., whether technologically more proficient employees will feel a better bargaining position in the company). Taking into account the specificity of a given organization, the model could also be used to predict employee attitudes towards the changing resource allocation due to AI and the dynamics of HR differentiation related to, for example, the introduction (or not) of a clear demarcation (creating mobility barriers) between AI-related and other positions.

Finally, since ‘there is a direct connection between fairness at work and in society at large’ (Rubery & Piasna, 2017, p. 53), policymakers can use the model as a support for planning decisions affecting the labor market. Awareness of the intra-organizational processes that shape and reinforce segmentation across companies and the labor market as a whole should make it possible to exert influence to reduce inequality effectively. For example, changes in labor law could be analysed not only in terms of their impact on management plans for the differentiated allocation of resources, but also in the perception of individual managers implementing the changes and employees themselves. A deeper understanding of intra-organizational processes combined with the analysis of labor market processes (e.g., employee transfers between companies as a result of changes in their relative position on the labor market) would allow for a full picture of the expected labor law changes and a reliable assessment of their consequences.