Introduction
Workers with reduced capacity have lower labour market participation compared to the population average (OECD, 2010; Eurostat, 2022). This has great social and economic consequences. From a social perspective, sustained labour market participation can facilitate economic independence, improved health, and social inclusion (van der Noordt et al., Reference van der Noordt, Ijzelenberg, Droomers and Proper2014; De Witte et al., Reference De Witte, Pienaar and De Cuyper2016). From an economic perspective, sustained labour market participation can limit social benefits expenditure while filling labour market shortages (Whitehead, Reference Whitehead2010; Clayton et al., Reference Clayton, Barr, Nylen, Burström, Thielen, Diderichsen, Dahl and Whitehead2012). Governments have tried to increase and sustain group level labour market participation through supply-side active labour market policies (ALMPs), which have aimed at increasing individual employability (OECD, 2010; McCollum, Reference McCollum2012; van der Aa and van Berkel, Reference van der Aa and van Berkel2014; Ingold and Stuart, Reference Ingold and Stuart2015; Frøyland et al., Reference Frøyland, Andreassen and Innvær2019). This has proved insufficient to increase and sustain group level labour market participation (Eichhorst and Konle-Seidl, Reference Eichhorst and Konle-Seidl2008; Kluve, Reference Kluve2010; Ingold and Stuart, Reference Ingold and Stuart2015; Lahey et al., Reference Lahey, Kirsch, MacDermid, Tompa and Gewurtz2019). As a result, public policy in European welfare states has shifted to demand-side policies (Böheim and Leoni, Reference Böheim and Leoni2018) that aim to engage employers to include workers with reduced capacity, and people from other vulnerable groups (McCollum, Reference McCollum2012; van der Aa and van Berkel, Reference van der Aa and van Berkel2014; Borghouts-van de Pas and Freese, Reference Borghouts-van de Pas and Freese2017; Böheim and Leoni, Reference Böheim and Leoni2018).
The implemented policies vary greatly between countries. Many have tried to directly regulate recruitment of people with disabilities through the implementation of quota, for which non-compliance is economically sanctioned (e.g. Germany, France, Austria, Belgium, Italy, Spain, and Poland) (Vornholt et al., Reference Vornholt, Villotti, Muschalla, Bauer, Colella, Zijlstra, Van Ruitenbeek, Uitdewilligen and Corbière2018). Governments in Scandinavia have tried to facilitate and incentivise recruitment and sustained employment for people from vulnerable groups by offering wage subsidies, training programs and support at the workplace (Gustafsson et al., Reference Gustafsson, Peralta and Danermark2013; Bredgaard and Halkjær, Reference Bredgaard and Halkjær2016; Castillo, Reference Castillo2019; Østerud, Reference Østerud2020; Østerud and Vedeler, Reference Østerud and Vedeler2022). Additionally, Norway has implemented a soft quota for state employers to increase the labour market participation of people from vulnerable groups (Østerud, Reference Østerud2020; Østerud and Vedeler, Reference Østerud and Vedeler2022). The Netherlands has combined direct regulation and incentives through requirements of job creation earmarked for people with disabilities, the implementation of quota, and by offering wage subsidies (Borghouts-van de Pas and Freese, Reference Borghouts-van de Pas and Freese2017; van Berkel, Reference van Berkel2021).
Despite a high number of implemented demand-side policies throughout Europe, the impact of the policies appears to be mixed. Studies on the effect of demand-side policies on employment opportunities for people remote from the labour market show that the policies have had low uptake with employers (Bredgaard and Halkjær, Reference Bredgaard and Halkjær2016; Bredgaard, Reference Bredgaard2018; Østerud, Reference Østerud2020; Østerud and Vedeler, Reference Østerud and Vedeler2022); found no association between implemented policies and probability of employment for people with disabilities (Reinders Folmer et al., Reference Reinders Folmer, Mascini and Van der Veen2020); or have been challenging to implement and realise in workplaces (Castillo, Reference Castillo2019; van Berkel, Reference van Berkel2021). This opens a discussion about the barriers to implementation, not only in terms of policy characteristics (Gustafsson et al., Reference Gustafsson, Peralta and Danermark2013; Bredgaard and Halkjær, Reference Bredgaard and Halkjær2016; Borghouts-van de Pas and Freese, Reference Borghouts-van de Pas and Freese2017; Vornholt et al., Reference Vornholt, Villotti, Muschalla, Bauer, Colella, Zijlstra, Van Ruitenbeek, Uitdewilligen and Corbière2018; Castillo, Reference Castillo2019; Østerud, Reference Østerud2020; van Berkel, Reference van Berkel2021; Østerud and Vedeler, Reference Østerud and Vedeler2022) but in terms of characteristics of employers that are relevant for effective implementation. More knowledge is needed on how demand-side policies can be implemented effectively to allow for a more targeted focus for where and how the policy impact and position of workers with reduced capacity can be improved. This article aims to gain these insights by studying the impact of employer characteristics on sustaining employment for workers with reduced capacity. Using comprehensive Norwegian register data from 2003-2017, this study aims to answer the following research question: How does employer size and extent of implemented disability policies impact sustaining employment for workers with reduced capacity?
Literature review
The scope of the literature on the impact of employer characteristics on sustained employment is limited. Most previous research has focused on how employer characteristics affect workers with reduced capacity in returning to work (RTW), while a few have studied how employer characteristics affect continued employment. For employer size, the evidence is inconsistent (Jansen et al., Reference Jansen, van Ooijen, Koning, Boot and Brouwer2021). Some studies found that working for a large employer is positively associated with RTW, retention, or continued employment (Hannerz et al., Reference Hannerz, Ferm, Poulsen, Pedersen and Andersen2012; Prang et al., Reference Prang, Bohensky, Smith and Collie2016; Schneider et al., Reference Schneider, Linder and Verheyen2016; van Ooijen et al., Reference van Ooijen, Koning, Boot and Brouwer2021); others found that working for a smaller employer is positively associated with early RTW or continued employment, or that small workplaces had lower entry into certified absence (Faucett et al., Reference Faucett, Blanc and Yelin2000; Krause et al., Reference Krause, Dasinger, Deegan, Rudolph and Brand2001; Holm et al., Reference Holm, Benn and Høgelund2007; Markussen et al., Reference Markussen, Røed, Røgeberg and Gaure2011); or no association was found at all (Cooper et al., Reference Cooper, Hankins, Rixon, Eaton and Grunfeld2013; Høgelund and Holm, Reference Høgelund and Holm2014; Bacon and Hoque, Reference Bacon and Hoque2022). The inconsistent evidence can probably be partially explained by different definitions of employer sizes between studies, as definitions can vary according to national standards. As a result, there is great variation in what is referred to as large and small employers.
Theoretically, large employers have resources to provide accommodations, formalised routines and practices, and often an HR-department, all of which can contribute positively to sustaining employment for workers with reduced capacity (Goss et al., Reference Goss, Goss and Adam-Smith2000; Borghouts-van de Pas and Freese, Reference Borghouts-van de Pas and Freese2017; Beatty et al., Reference Beatty, Baldridge, Boehm, Kulkarni and Colella2019; Hyggen and Vedeler, Reference Hyggen and Vedeler2021; Bacon and Hoque, Reference Bacon and Hoque2022). This is, however, the opposite of what Stone and Colella (Reference Stone and Colella1996) predicted in their model of factors affecting treatment of people with disabilities in workplaces. They highlight that workplaces with bureaucratic and standardised procedures can be disadvantageous for people with disabilities, because people with disabilities might not be able to comply with the set standards and rules (Stone and Colella, Reference Stone and Colella1996). Others contend that small employers provide benefits of high informality, personalised treatment, more flexibility in terms of working hours and tasks, and greater job autonomy because of few bureaucratic obstacles (Tsai et al., Reference Tsai, Sengupta and Edwards2007; Storey et al., Reference Storey, Saridakis, Sen-Gupta, Edwards and Blackburn2010; Harney and Alkhalaf, Reference Harney and Alkhalaf2021; Bacon and Hoque, Reference Bacon and Hoque2022). These are factors that can facilitate personalised and flexible adjustment that might be necessary for workers with reduced capacity to sustain employment. As a result, there are theoretical reasons to expect both smaller and larger employers to contribute positively to sustaining employment for workers with reduced capacity.
At the same time, the inconsistent evidence on the effect of employer size on RTW or continued employment could be partially explained by other factors not considered in previous research. Several studies have found that extent of implemented disability policies is associated with RTW and continued employment (Williams and Westmorland, Reference Williams and Westmorland2002; Gensby et al., Reference Gensby, Labriola, Irvin, Amick and Lund2014; Jakobsen and Lillefjell, Reference Jakobsen and Lillefjell2014; Amick et al., Reference Amick, Lee, Hogg-Johnson, Katz, Brouwer, Franche and Bültmann2017; Beatty et al., Reference Beatty, Baldridge, Boehm, Kulkarni and Colella2019; van Ooijen et al., Reference van Ooijen, Koning, Boot and Brouwer2021). Disability policies often come in terms of practices, routines, and guidelines for how to hire, retain and accommodate workers with reduced capacity (Beatty et al., Reference Beatty, Baldridge, Boehm, Kulkarni and Colella2019). Employer specific policies can be formulated from official (governmental) regulations (e.g. anti-discrimination laws), or organisational goals or missions related to inclusion and treatment of workers with reduced capacity. The degree of policy driven regulation often increases with employer size (Storey et al., Reference Storey, Saridakis, Sen-Gupta, Edwards and Blackburn2010; Bacon and Hoque, Reference Bacon and Hoque2022) to meet the demands of a more complex organisation. Following this, large employers can have a higher extent of implemented disability policies (Chan et al., Reference Chan, Tansey, Iwanaga, Bezyak, Wehman, Phillips, Strauser and Anderson2021; Bacon and Hoque, Reference Bacon and Hoque2022). At the same time, smaller employers, though maybe lacking the same extent of implemented (disability) policies, can develop ad hoc policies and practices when the question for how to accommodate and retain workers with reduced capacity arises (Harney and Dundon, Reference Harney and Dundon2006; Storey et al., Reference Storey, Saridakis, Sen-Gupta, Edwards and Blackburn2010; Lindsay et al., Reference Lindsay, Cagliostro, Leck, Shen and Stinson2019). Extent of implemented disability policies could therefore partially explain the inconsistent evidence on the effect of employer size on RTW, continued employment, and sustained employment for workers with reduced capacity.
Examining how employer size and extent of implemented disability policies impacts probability of sustaining employment for workers with reduced capacity contributes to the existing literature in two ways. Firstly, it can help clarify the relationship between employer size and extent of implemented disability policies and probability of sustaining employment for workers with reduced capacity. Secondly, it can contribute to understanding under which labour market conditions workers with reduced capacity face the lowest and highest chances of sustaining employment. Identifying these conditions can help guide policymakers in which employers to target when designing and implementing demand-side policies. Demand-side policies that target employers displaying characteristics that negatively impact sustaining employment can contribute to more effectively increasing group level labour market participation and sustained employment for workers with reduced capacity.
Register data is well suited to study the impact of employer characteristics on probability of sustaining employment, as the coverage of almost two decades allows for the consideration of complete employment records with the same employer after entry to a disability benefits system. With these records I construct a proxy measure for extent of implemented disability policies. The measure assumes that employers where workers with reduced capacity have a high mean employment duration after entry to disability benefits are better at accommodating and facilitating the workers, indicating a higher extent of implemented disability policies. Additionally, I consider how industry, sector, and workforce composition in terms of gender, age, mean wage, and education within employers impact the probability of sustaining employment for workers with reduced capacity.
The Norwegian case
Norwegian demand-side policy differs from the direct regulation of central-European policy. In 2001, Norway implemented the Inclusive Work Life agreement that sought to reduce sick leave absence and improve work retention for people with ill health, disabilities, or reduced work capacity (Mandal and Ose, Reference Mandal and Ose2015; Hyggen and Vedeler, Reference Hyggen and Vedeler2021). The agreement was voluntary until 2019, after which participation in the agreement became mandatory for both public and private employers. In 2017, government launched a new policy agenda, the ‘Joint Inclusion Effort’ (Inkluderingsdugnaden), that introduced a soft quota for state employers, requiring 5 per cent of all new recruitments to be made from vulnerable groups (Østerud, Reference Østerud2020; Østerud and Vedeler, Reference Østerud and Vedeler2022). Non-compliance is, however, not sanctioned, and no regulations were imposed on private employers. Additionally, Norwegian government has offered extensive support to employers to increase inclusion and retention for workers with reduced capacity (Aakvik and Dahl, Reference Aakvik and Dahl2006; Clayton et al., Reference Clayton, Barr, Nylen, Burström, Thielen, Diderichsen, Dahl and Whitehead2012).
Despite Norway’s extensive policy interventions, a disability employment gap of 34.5 per cent remains, which is 12 percentage points higher than the EU average (Eurostat, 2022). In a study on employer participation in ALMPs, Hyggen and Vedeler (Reference Hyggen and Vedeler2021) found no association between employers participating in the Inclusive Work Life agreement and probability to hire young people remote from the labour market. In studies on the uptake of the ‘Joint Inclusion Effort’ with state employers, Østerud (Reference Østerud2020) found that only 3.1 per cent fulfil the employment quota. Østerud and Vedeler (Reference Østerud and Vedeler2022) found that state employers remain passive in hiring, despite having positive attitudes towards the soft quota. As the implemented policies have been less than effective, more knowledge is needed on what impact employer characteristics have on sustaining employment for workers with reduced capacity.
Methods
Data
I used Norwegian population register data from 2003-2017 obtained from Statistics Norway. Norwegian register data is generally considered to be of high quality. An evaluation of employment register data at the micro level shows that the registers systematically overestimate employment rates, but that the agreement between survey data and register data on employment rates is high (Villund, Reference Villund2010). The dataset used for analysis was compiled from multiple registers from Statistics Norway, on linked employer-employee (LEE) records, disability benefit records, demographic records, and educational records.
Sample
The sample of workers with reduced capacity consists of workers that enter disability benefits at different timepoints. The sample was identified by extracting the first record of entry to a disability benefits system between 2003 and 2017. Entry records to disability benefits were extracted from four different disability benefit registers, consisting of medical rehabilitation benefits, temporary disability benefits, work assessment allowance (AAP, replacing the two former benefits as of 2010) and disability pensions. The three first benefits are temporary disability benefits systems. For eligibility, work capacity must be reduced by at least 50 per cent due to injury or ill health. For work-related injury, reduced work capacity of 30 per cent is sufficient. Additionally, work capacity of AAP recipients must be improvable, either through treatment, accommodated work, or through counselling from Norwegian Labour and Welfare administration (NAV). Disability pensions are permanent benefits for individuals with permanently reduced work capacity. The inclusion criterion for the analysis was that the individuals with reduced work capacity had a job at the time of entry to disability benefits. Individuals that had no residual work capacity and under eighteen years of age were excluded. This led to a sample of 46 306 observations (single episode data), that were followed for the duration of the employment record after entry to disability benefits. The sample consisted of two groups of workers with reduced capacity: one group entered directly to a disability benefits system (n=6690), while the other group had been on sick leave before entry (n=34896). The sick leave registers did not contain data for 2017, for which group membership is undetermined (n=4720). For a linear probability model (LPM) that predicts group membership, see supplementary material S1.
Dependent variable
The dependent variable, employment ending, is binary and defined as whether employment ends in the observed month after entry to disability benefits (1) or not (0). Observations for which employment does not end in the entirety of the observation period are right-censored. A long employment period after entry to disability benefits indicates sustained employment.
Independent variables
Employer characteristics
I used the LEE-records to construct aggregated employer characteristics from individual-level variables, which are time invariant within employers, within years. For employer size, three categories were distinguished following the EU definition: small, medium, and large employers (see table 1) (EU Commission, 2003). Micro (<10 employees) and small (>9 and <50 employees) employers were combined into one category due to a small sample of micro employers (n=1558). The proxy measure for extent of implemented disability policies was inspired by the proxy measure used by van Ooijen et al. (Reference van Ooijen, Koning, Boot and Brouwer2021). I defined the proxy measure as the mean employment duration in months of co-workers with reduced capacity after entry to disability benefits within the same employer, in the same year (mean=25.88, SD=28.02). Mean employment duration after entry to disability benefits can indicate the degree to which employers are able to accommodate and retain workers with reduced capacity. This is likely reflected in implemented disability policies and practices within employers. The measure therefore assumes that employers with a high mean employment duration for co-workers with reduced capacity have a higher extent of implemented disability policies. The measure is restricted to co-workers to avoid individual employment duration becoming predictive of individual sustained employment. As a result, only employers with at least two workers with reduced capacity are included in the sample. The measure was divided into quartiles. The first quartile ranges from one to six months, the second quartile from six to eighteen months, the third quartile from eighteen to thirty-four months, and the highest quartile from thirty-four to 180 months. Age composition was defined as the fraction of workers over forty-five years of age. Gender composition was defined as the fraction of female employees. Education composition was defined as the fraction of workers with higher education (short university degree or long university degree). Wage composition was defined as the mean wage of all employees, within employers. The composition variables were split into quartiles. Observations with missing information on the aggregated measures were excluded within employers. Industry and sector were retrieved from employment registers. Industry was based on SN2002, which distinguishes eighteen categories for industry. Categories were combined to reduce model complexity, resulting in seven industry categories (see table 1). Sector distinguished between public (0) and private (1) employers. Observations on industry and sector with missing information were excluded.
Table 1 Summary statistics
Event = Percentage of sample that experienced (1) employment ending.
Individual characteristics
Individual characteristics were included to control for the impact of individual characteristics on sustaining employment for workers with reduced capacity. Sex (binary), age (continuous), education (binary, 0 = no university degree, 1 = short or long university degree), marital status (binary), wage (quartiles), and time on disability benefits (continuous) were included in the analysis. All variables were defined in terms of the value of the observation at the time of entry to disability benefits. Observations with missing information on individual characteristics were excluded.
Analytical strategy
To evaluate how employer characteristics affect sustaining employment for workers with reduced capacity, data was analysed using parametric proportional hazard (PH) models with gamma distributed shared frailty at the employer level (Gutierrez, Reference Gutierrez2002). Shared frailty models are models in which frailties (latent multiplicative effects on the hazard function) are shared among groups of individuals (Gutierrez, Reference Gutierrez2002). In the case of this study, frailty is shared among workers with reduced capacity within the same employer. The models therefore allow for estimating the risk of employment ending while accounting for intragroup correlation. Prior to model estimation, univariate analysis of all variables was conducted. For p-values lower than 0.25, variables were included. Categorical variables were evaluated in terms of log-rank tests and proportionality of Kaplan Meier curves. The proportionality assumption held visually, and all tests of equality between categories were significant. For continuous variables (age and time on disability benefits), p-values were obtained through proportional cox models. All variables had p-values below the 0.25 threshold. Parametric distribution was chosen based on model comparison of AIC and BIC statistics and Csnell pseudoresiduals (Blossfeld et al., Reference Blossfeld, Rohwer, Schneider, Blossfeld, Rohwer and Schneider2019: 226). The Lognormal distribution barely outperformed the Gompertz distribution in AIC and BIC statistics. The Gompertz distribution performed best through visual inspection of Csnell pseudoresiduals. Based on this, the Gompertz distribution was selected for analysis. For details on selection of parametric distribution, see supplementary material S2.
After selecting parametric distribution, backwards selection was performed to identify variables that did not contribute to the model. The individual characteristics immigration status and socio-demographic background, defined as highest parental education, were excluded. After identifying the final model for analysis, two main models were estimated: one model with employer size, and one with employer size and extent of implemented disability policies. Additionally, a model with an interaction between employer size and extent of implemented disability policies was estimated. The model was estimated as a robustness test to investigate whether the impact of extent of implemented disability policies on probability of sustaining employment varies with employer size. A limitation of using parametric hazard models with shared frailty is that with passing time the effect of the frailty on the population hazard increases while the effect of the covariates decreases (Gutierrez, Reference Gutierrez2002). With time, therefore, the effect of the covariates (e.g. employer size) becomes less of a factor and shared frailty within employers becomes more of a factor in determining the risk of employment ending (Gutierrez, Reference Gutierrez2002). Each parametric shared frailty model is therefore complemented with a proportional cox model to capture duration effects of employment without shared frailty.
Results
Table 1 presents descriptive statistics for the sample of workers with reduced capacity. 55.6 per cent (n=25 737) of workers with reduced capacity experience employment ending in the observation period. The average employer is in the private sector, in the household activities and other services industry, and is of medium size. For workforce composition in terms of gender, age, mean wage, and education, the percentage of workers experiencing employment ending (1) decreases from the lowest to the highest quartiles. The average worker with reduced capacity is female, forty years of age (SD 9.23), unmarried, has no university degree, and has a mean of twenty-one months on disability benefits (SD 20.40).
Table 2 presents the results of the estimated parametric Gompertz models with gamma distributed shared frailty at the employer level. For employee characteristics and year effects, see supplementary material S3. In Model 1a, the results show that workers with reduced capacity face the lowest risk of employment ending while working for small employers. Working for a medium or large employer significantly increases risk of employment ending, compared to small employers. Risk of employment ending increases with 12 and 21 per cent, respectively. In model 2a, similar effects are found for employer size. For extent of implemented disability policies, working for an employer with extent of implemented disability policies in the second, third, or highest quartile significantly reduces risk of employment ending, compared to the first quartile. Risk of employment ending is reduced with 8, 16, and 23.6 per cent, respectively. The interaction between employer size and extent of implemented disability policies yielded no significant results, which means that the impact of extent of implemented disability policies on risk of employment ending does not vary by employer size. See supplementary material S4 for the results of the interaction.
Table 2 Estimation results
Hr = hazard rates. Se = standard errors. *p = .05, **p = .01, ***p = .001.
Model 1a & 2a = parametric Gompertz models with employer shared frailty.
Model 1b & 2b = proportional cox models.
For the proportional cox models (model 1b and 2b), similar results are found. In model 1b, working for a medium or large employer increases risk of employment ending relative to small employers. The model predicts a slightly lower hazard rate for large employers, compared to the parametric specification. For model 2b, the proportional cox model again predicts slightly lower hazard rates for employer size compared to the parametric specification. For extent of implemented disability policies, the predicted risks are in the same direction as for the parametric specification. The predicted reductions in risk differ slightly between models. The proportional cox model predicts a lower reduction in risk for the second quartile, and higher reductions in risk for the third and highest quartile, compared to the parametric specification.
Figure 1 to 3 presents the predicted population survival functions for workers with reduced capacity by employer size for quartiles of extent of implemented disability policies, with all other predictors held at their mean. The figures present survival functions that are unconditional on the frailty. The survival functions decrease with time, which means that the probability of sustaining employment decreases with time. The survival functions do not reach zero at the end of analysis time (t=180), which means the predicted probability for survival is fairly high. The survival functions decrease steeply at the beginning of analysis time, after which the decrease levels out around analysis time equal to 100. This means that the probability of sustaining employment with time decreases at a lower rate. For all employer sizes, extent of implemented disability policies in the lowest quartile predicts the lowest probability of sustaining employment, while extent of implemented disability policies in the highest quartile predicts the highest probability of sustaining employment. Similar to the estimation results in table 2, working for small employers (figure 1) predicts the highest probabilities of sustaining employment, while working for large employers (figure 3) predicts the lowest probability of sustaining employment, for all quartiles of extent of implemented disability policies. The difference in predicted probability of sustaining employment between each quartile of extent of implemented disability policies is lowest for medium employers (figure 2), with about 2 percentage points between each quartile. The difference in predicted probability of sustaining employment between extent of implemented disability policies in the highest and lowest quartile at the end of analysis time is 11 percentage points for small and large employers, and 8 percentage points for medium employers.
Figure 1. Predicted population survival functions for small employers, for quartiles of extent of implemented disability policies.
Figure 2. Predicted population survival functions for medium employers, for quartiles of extent of implemented disability policies.
Figure 3. Predicted population survival functions for large employers, for quartiles of extent of implemented disability policies.
Discussion
This article has investigated how employer size and extent of implemented disability policies impact the probability of sustaining employment for workers with reduced capacity. The results show that working for medium or large employers negatively impacts the probability of sustaining employment for workers with reduced capacity, relative to small employers. The negative impact on probability of sustaining employment gradually increases with employer size. This finding is inconsistent with some previous research, as several studies have found positive associations between large employers and RTW or continued employment (Hannerz et al., Reference Hannerz, Ferm, Poulsen, Pedersen and Andersen2012; Prang et al., Reference Prang, Bohensky, Smith and Collie2016; Schneider et al., Reference Schneider, Linder and Verheyen2016; van Ooijen et al., Reference van Ooijen, Koning, Boot and Brouwer2021). Based on the results, small employers with low standardisation, few bureaucratic obstacles, and greater opportunity for personalised treatment provide the best conditions for sustaining employment for workers with reduced capacity. This is supported theoretically (Stone and Colella, Reference Stone and Colella1996; Krause et al., Reference Krause, Dasinger, Deegan, Rudolph and Brand2001; Holm et al., Reference Holm, Benn and Høgelund2007; Tsai et al., Reference Tsai, Sengupta and Edwards2007; Storey et al., Reference Storey, Saridakis, Sen-Gupta, Edwards and Blackburn2010; Harney and Alkhalaf, Reference Harney and Alkhalaf2021; Bacon and Hoque, Reference Bacon and Hoque2022) and empirically in a few studies on early RTW, continued employment, and entry to certified absence (Faucett et al., Reference Faucett, Blanc and Yelin2000; Krause et al., Reference Krause, Dasinger, Deegan, Rudolph and Brand2001; Holm et al., Reference Holm, Benn and Høgelund2007; Markussen et al., Reference Markussen, Røed, Røgeberg and Gaure2011). Extent of implemented disability policies could not partially explain this effect. This raises the question of whether formal organisation, practices, and resources are key determinants of effectiveness in sustaining workers with reduced capacity, as has classically been argued (Goss et al., Reference Goss, Goss and Adam-Smith2000; Borghouts-van de Pas and Freese, Reference Borghouts-van de Pas and Freese2017; Beatty et al., Reference Beatty, Baldridge, Boehm, Kulkarni and Colella2019; Bacon and Hoque, Reference Bacon and Hoque2022). At the same time, Jansen et al. (Reference Jansen, van Ooijen, Koning, Boot and Brouwer2021) found that the evidence on the association between employer size and RTW or continued employment is inconsistent. Following this, more research is required to determine the impact of employer size on the probability of sustaining employment for workers with reduced capacity, and whether other factors not considered in previous research can partially explain the inconsistent evidence.
In line with previous research (Williams and Westmorland, Reference Williams and Westmorland2002; Gensby et al., Reference Gensby, Labriola, Irvin, Amick and Lund2014; Jakobsen and Lillefjell, Reference Jakobsen and Lillefjell2014; Amick et al., Reference Amick, Lee, Hogg-Johnson, Katz, Brouwer, Franche and Bültmann2017; Beatty et al., Reference Beatty, Baldridge, Boehm, Kulkarni and Colella2019; van Ooijen et al., Reference van Ooijen, Koning, Boot and Brouwer2021), I find that a higher extent of implemented disability policies positively impacts the probability of sustaining employment for workers with reduced capacity. The positive impact on probability of sustaining employment gradually increases with each quartile of extent of implemented disability policies. Additionally, the interaction between employer size and extent of implemented disability policies yielded no significant results, which means that a higher extent of implemented disability policies contributes positively to probability of sustaining employment for workers with reduced capacity, irrespective of employer size. At the same time, the proxy measure for extent of implemented disability policies yields no information on what types of policies have been implemented by employers. In a systematic review on employer provided disability management programs, Gensby et al. (Reference Gensby, Labriola, Irvin, Amick and Lund2014) found that it was not possible to determine which programs or program components drove the effectiveness of the implemented policies, and that in-house disability management programs might differ between employers based on size. This opens a discussion on whether types of implemented policies and practices differ according to employer size, how they differ, and whether the policies vary in effectiveness. To further develop the measure of extent of implemented disability policies, more knowledge is needed to determine the impact of specific disability policies and practices on sustaining employment for workers with reduced capacity, and whether the policies differ in implementation and effectiveness according to employer size.
For the other employer characteristics estimated, impact of workforce composition in terms of workers over forty-five years of age, balanced gender composition, and high proportion of employees with higher education positively impacts probability of sustaining employment for workers with reduced capacity. This is in line with previous research (Markussen et al., Reference Markussen, Røed, Røgeberg and Gaure2011; van Ooijen et al., Reference van Ooijen, Koning, Boot and Brouwer2021). The results indicate that employers that have a diverse workforce in terms of gender, age, and high proportion of employees with higher education provide better conditions for sustaining employment for workers with reduced capacity, likely because the employers have experience in sustaining various groups of workers. Working for private sector employers positively impacts the probability of sustaining employment for workers with reduced capacity, relative to public employers. This means that public employers have worse conditions for sustaining workers with reduced capacity compared to their private counterparts. As Østerud (Reference Østerud2020) found, state employers were only passively participating in the Joint Inclusion Effort, for which the employers were facing contradicting demands from the soft quota on the one hand, and productivity demands and cost reductions on the other. Such high demands on efficiency in specialised organisations can give public employers little leeway to integrate workers with reduced capacity. This is problematic, as public employers should be at the forefront in sustaining workers with reduced capacity as organisational representatives of the government that designs and implements both supply- and demand-side policies.
The results indicate three main lines of inquiry for future research. Firstly, more knowledge is needed to determine the impact of employer size in sustaining workers with reduced capacity. If future research finds that small employers provide the best conditions for sustaining employment for workers with reduced capacity, studying how these characteristics can be adapted and transferred to larger employers can help increase group level labour market participation and sustained employment. Otherwise, more research is needed to identify factors that can explain the inconsistent evidence of employer size on RTW, continued employment, and sustained employment.
Second, to further investigate the effect of extent of implemented disability polices on probability for sustaining employment, future research should aim to develop a measure linked to implemented policies and practices. This could increase validity of the measure while allowing for a more fine-grained distinction between employers based on hiring behaviour, and what types of policies have been implemented. Previous research has developed typologies that differentiate between engaged and non-engaged employers based on participation in and attitudes toward ALMPs (Bredgaard, Reference Bredgaard2018) and hiring behaviour (Hemphill and Kulik, Reference Hemphill and Kulik2016). The typologies are useful for mapping non-engaged employers that can be converted to engaged employers (e.g. non-hirers and light hirers; passive employers and dismissive employers) (Hemphill and Kulik, Reference Hemphill and Kulik2016; Bredgaard, Reference Bredgaard2018), as these are employers in which the position of workers with reduced capacity can be improved. Designing a typology that takes extent of implemented disability policies into account could help identifying which non-engaged employers have policies and practices that are conducive to inclusion and sustained employment for workers with reduced capacity, which could provide a basis for which employers to target with demand-side policies. Additionally, a measure linked to implemented disability policies and practices would contribute to the knowledge of which policies and practices positively impact sustaining employment for workers with reduced capacity, and whether the policies and their effectiveness differ between employer sizes. Such a measure also minimises individual selection effects, which the proxy measure in this study cannot control for: workers with reduced capacity might be more likely to stay with the current employer if reduced work capacity limits other employment options, which could partly explain long employment duration after entry to disability benefits.
Lastly, there is inconsistent evidence for the association between sector and sustaining employment (Jansen et al., Reference Jansen, van Ooijen, Koning, Boot and Brouwer2021) for workers with reduced capacity. Future research should strive to clarify how public and private employers impact probability of sustaining employment with respect to employer size, focusing on similarities and differences in demands on productivity, cost efficiency, and recruitment and retention procedures. Only by equal efforts of both public and private organisations can higher labour market participation and sustained employment for workers with reduced capacity be achieved.
This study has focused on single episode data and the duration of single employment records. As such, sustaining employment in this study is only related to probability of sustaining employment with the same employer after entry to disability benefits. To better understand how employers contribute to sustaining employment for workers with reduced capacity, future research could define sustaining employment as sustaining labour market participation and consider multiple (un)employment records over time. This approach could contribute to better understanding individual employability and how employers contribute to it, while accounting for more complete labour market trajectories. This approach opens for differentiating between workers whose employment records end because they have found new jobs with different employers and workers that become unemployed, which the current study does not. The data in this study does not contain health or diagnostic information, only degree of reduced capacity for records from disability pensions. As such, I cannot control for degree of reduced capacity. It is unlikely, however, that an individual with no residual work capacity has been included in the analysis, as the sample consists of employed individuals only. Additionally, I control for time on disability benefits, which can capture some of the variation in degree of reduced capacity: the more severe condition, health, or injury, the longer the period on disability benefits.
Norway has implemented demand-side policies focused on increasing labour market participation for workers with reduced capacity (and other vulnerable groups) (Aakvik and Dahl, Reference Aakvik and Dahl2006; Clayton et al., Reference Clayton, Barr, Nylen, Burström, Thielen, Diderichsen, Dahl and Whitehead2012; Mandal and Ose, Reference Mandal and Ose2015; Østerud, Reference Østerud2020; Hyggen and Vedeler, Reference Hyggen and Vedeler2021; Østerud and Vedeler, Reference Østerud and Vedeler2022). Despite this, labour market participation for this group remains lower compared to the population average (Eurostat, 2022). As the results show, sustaining employment is impacted by and varies with employer characteristics in terms of size and extent of implemented disability policies. This study suggests that policymakers in Norway should consider both employer size and extent of implemented disability policies when designing new policies. This consideration could help policymakers to better target employers that have low levels of implemented disability policies, where intervention would be most impactful. Such an approach could be extended to other countries that implement demand-side policies and face similar challenges to sustain workers with labour restrictions.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/S1474746423000027.
Financial support
This research was financed by the Norwegian Research Council, grant number 301045.
Open access
