Disability and labor market outcomes

Measuring disability

While the availability of comparable international survey data such as that presented in Figure 1 and Figure 2 appear to provide opportunities for cross-country analysis, there are important measurement issues involved. The magnitude and nature of international variation, particularly in terms of disability prevalence, raise important concerns about the extent to which self-reported disability, which depends on the social, economic and policy context, is comparable across countries [1]. Indeed, the incentives to self-report disability may depend on social acceptability and financial implications, which may relate to country specific institutional features, such as the welfare system and anti-discrimination legislation. Nevertheless, some common patterns have been observed: rates of disability are typically higher in northern than in southern Europe, these rates increase with age and decrease with more formal educational qualifications. Across the EU, for example, the percentage of the population reporting disability among those aged 55–64 (26%) is eight times higher than among those aged 15–24 (3%).

The majority of evidence used in this article relies on self-reported measures of disability, which are now routinely available from international survey data. They have, however, been subject to a number of criticisms, and studies have sought to explore their validity using more objective information such as activity restrictions arising from functional limitations. But these objective measures of health are also likely to suffer from measurement error as the concept of disability will itself depend on the social and economic environment. As such, rather than using an alternative measure in place of self-reported information, studies examined how self-reported disability varies compared to “true” disability that is, for example, constructed from objective health measures or receipt of disability benefits. The results are, however, mixed and inconclusive, with studies finding evidence both for and against the use of self-reported disability data.

The analysis of a subset of disabled individuals, who receive disability welfare payments, forms a largely separate strand of literature. While this is arguably a more objective measure of disability, in the sense that recipients typically have to meet specified medical criteria, eligibility for, and therefore receipt of benefits depends on the nature of the scheme. Further, despite institutional differences, the majority of these schemes are designed as income replacement, and therefore tend to impose intentional and substantial restrictions on “permitted employment”. This design feature limits the usefulness of disability, when defined in relation to welfare benefit, in analyzing individual labor market outcomes. Nevertheless, cross-country variation in receipt of disability benefits among older workers, which substantially exceeds variation in indicators of objective health, is sufficient to suggest that disability welfare forms a route into early retirement in some countries. Moreover, country-specific studies, such as those based on changing benefit regimes, provide important evidence of a causal relationship between the level of disability benefit and non-participation in the labor market. As such, the design of the disability welfare system is undoubtedly an important contributory factor to the broader self-reported disability employment gap.

The nature of disability welfare schemes have attracted increasing attention, at least partially due to significant growth in disability benefit caseloads and the associated financial pressure, particularly in parts of northern Europe, the US, the UK, and Australia. This growth has occurred over a period where objective measures of health have generally been improving and dominant explanations for growth instead relate to the design of the scheme (e.g. relaxation in eligibility requirements and increasing relative generosity), changes in demographics, female labor force participation and reduced demand for low skilled workers [2]. Recent reforms of disability benefit systems have tended to contain active strategies to encourage re-engagement with work designed to enhance the (typically low) rate of exit from disability benefits. Tighter medical (among other) eligibility criteria have also been introduced to reduce the inflow of recipients and to better target support to those that are unable to work. While there is recognition of the difficulty associated with attempting to achieve two conflicting goals, that is, providing financial support to those unable to work while at the same time encouraging those who can to retain or re-engage employment, there has been some recent success, at least in terms of reducing caseloads, particularly in the Netherlands. Nevertheless, to understand the broader disability employment gap, future work needs to examine the extent to which such reforms have led to continued labor market attachment (or reattachment) rather than benefit displacement.

Disability, employment and earnings

The size of the employment gap (see Figure 2), combined with its persistence over time and across countries, has motivated a body of evidence which attempts to identify the drivers of disability labor market inequality and monitor its trend over time. The latter, in particular, has been used to assess the effectiveness of major changes in policy and legislation. This evidence frequently considers hourly labor market earnings, where the disability gap is significant but often more modest, generally being between 10 and 20%.

Explanations for the disability-related employment gap vary and include: pre-existing disadvantage, changes in capacity for and ability to work, and changes in preferences for work, such as those arising from changes in the value of leisure and/or eligibility for welfare support. They also include reverse causality, including justification bias; that is, the incentive for those who are out of work to legitimize their situation by subsequently reporting disability. A key issue has, however, been the influence of discrimination or unequal treatment by employers arising from prejudice or imperfect information (whereby the employer uses disability as a signal of low productivity). Studies have attempted to distinguish discrimination from the disadvantage associated with other personal and work-related characteristics. This type of analysis asks to what extent gaps in the raw data reflect disability, per se, rather than other factors, such as age and education, which are correlated with disability. A substantial proportion of both the employment and earnings gaps are found to relate to disability, or what is often referred to as being unexplained by other factors. In the UK, for example, about 75% of the employment gap, and between 50–75% of the earnings gap, are found to relate to disability [3], [4].

One main limitation of this type of analysis is that it is difficult to control for other unobserved factors such as the impact of disability on productivity at work or preferences to work. As such, the unexplained gap is almost certainly an overestimate of disability discrimination. Studies have attempted to tackle this issue by controlling for functional limitations and/or by using different definitions of disability to identify groups of disabled individuals who are more or less likely to experience discrimination or productivity reductions at work. These studies tend to find that discrimination plays a less important role [3], [4], [5]. Nevertheless, contributions using an alternative approach based on correspondence studies, in which job applications of disabled and non-disabled people with otherwise identical CVs are sent to employers in response to a job advert, find that rates of invitation to interview are significantly lower for disabled relative to non-disabled applicants, consistent with employer discrimination.

Studies have sought to evaluate the impact of major changes in legislation which have made discrimination against disabled individuals unlawful in several countries, including the 1990 Americans with Disabilities Act (ADA) in the US and the 1995 Disability Discrimination Act (DDA) in the UK, by comparing the outcomes of disabled and non-disabled individuals before and after the introduction of the legislation. Both pieces of legislation contain two main components: an antidiscrimination element that makes disability discrimination unlawful, and a reasonable adjustment element that requires employers to make changes to the workplace and work practices to prevent a disabled person from being disadvantaged. Although the threat of legal action related to disability discrimination on hiring would be expected to increase the employment of disabled individuals, the anticipated increase in firing costs arising from wrongful termination combined with the costs of accommodation are predicted to act in the opposite direction. It is the latter that is anticipated to dominate and, due to the expected increased costs for employers when hiring disabled individuals, is predicted to reduce demand for disabled workers [6].

Overall, there is little evidence of positive employment effects arising from the introduction of such legislation [6], [7]. Moreover, negative employment effects in the US have been found to vary by firm size and by variations in disability discrimination charges among states, which is consistent with an adverse influence of the ADA [6]. Indeed, when using variation in pre-existing legislation between US states, there is preliminary evidence that it was the introduction of the reasonable accommodation element of the legislation that had short-run negative consequences [8]. Nevertheless, these findings have not gone undisputed, with factors other than the ADA—for instance, the economic cycle and changes in the disability welfare regime—put forward as alternative explanations for the decline in the employment rate among disabled individuals in the US.

Disability and disadvantage in work

Recent studies have considered a broader range of labor market outcomes including hours of work and the nature of employment. The concentration of disabled workers in part-time and self-employment raises questions about the extent to which this reflects “push” factors, such as inequality of treatment, or “pull” factors, including the ability to accommodate disability in work. Such analysis has also started to consider the experience of work using subjective measures of skill utilization, job satisfaction, perceptions of managers and employee commitment. Relative to their non-disabled counterparts, disabled workers tend to report more negative experiences across a range of in-work outcomes; this trend is evident across several countries including the US, Spain, the UK, and Australia [9]. Further, this is not explained by differences in personal characteristics or more objective work-related characteristics, such as hours worked or occupation, and therefore exists, on average, between comparable disabled and non-disabled workers in comparable jobs. Accordingly, differences in work-related wellbeing are consistent with higher rates of reporting of bullying and harassment from employers and co-workers among disabled relative to non-disabled employees in the UK.

An interesting question, which can be explored using matched employee–employer data, is the role of the employer and the influence of specific workplace policies and practices on this disability disadvantage. While these issues remain underexplored, recent US evidence finds that the disability gap in perceptions disappears in workplaces that are viewed as the most fair among all employees, pointing to the importance of “corporate culture” [9]. Understanding the work-related wellbeing of disabled workers is not only important in its own right, but also because of its likely contribution to the employment and earnings gaps via the impact on the recruitment, retention and productivity of disabled individuals.

Longitudinal evidence

A major criticism of the literature is the focus on cross-sectional data and associations/correlations between variables rather than causal relationships. More recently, longitudinal evidence, which is able exploit the dynamic nature of disability to track the same individual over time has been used to identify the disadvantage associated with disability measured relative to the same individual pre-onset (i.e. before that individual became disabled), rather than a similar non-disabled individual, who may differ in a range of unobserved ways. Among other things, such analysis is able to separate the disadvantage associated with disability onset from pre-existing disadvantage, and is able to use the timing of disability onset relative to the observed disadvantage to rule out reverse causality. Longitudinal evidence has one further advantage: it is able to identify and distinguish between disadvantage associated with different dynamic patterns of disability, particularly the duration of disability. Indeed, analysis of the dynamics of disability highlights that, for many, disability is not permanent.

Although much of the existing longitudinal evidence is based on US data, there have also been important recent contributions for Germany, the UK, and Australia. Several key findings emerge from this literature. First, there is evidence that disability onset is associated with employment and earnings disadvantages relative to the same individual pre-onset, which is consistent with a causal explanation. Further, the dynamics of disability are important: those with chronic disability, which is defined as persisting post-onset, experience greater disadvantage at onset and, in contrast to arguments that individuals adapt, this disadvantage is exacerbated post-onset. Finally, self-reported severity is a key driver of the magnitude of disadvantage. For example, those who report chronic severe disability experience more than 3.5 times the reduction in annual working hours ten years into the post-onset period [10]. Further, this type of framework has been used to consider the broader impact of disability on wellbeing, recognizing that the implications of changes in individual labor market status may have a less pronounced impact on household income and/or consumption when there is support within the household or from the government, such as disability benefit income. Indeed, recent evidence on the negative impact of disability onset on subjective self-reported life satisfaction raises interesting questions for policymakers about how disadvantage should be measured.

The focus on the dynamics of disability has also raised questions about the influence of the timing of onset [11]. It is important to distinguish between those who are disabled at birth or during childhood and those who have already entered the labor market when becoming disabled because the barriers to employment for these two groups may differ. Among the first group, disability may affect the accumulation of human capital and will precede entry into the labor market, whereas human capital is likely to have been largely determined prior to disability onset among the latter, where the key issue may instead be the retention of employment [12]. Indeed, in [11] a distinction is made between general human capital, which is valued equally for disabled and non-disabled individuals (such as formal education or training); healthy human capital (that is, human capital that cannot be utilized due to disability), which is valued only for non-disabled individuals; and disability human capital, which is valued for those with disability (such as learning to use adaptations). If healthy human capital increases with age, those with age-onset disability will face more severe disadvantage at onset. Further, those who are disabled at a younger age should have more incentive to invest in disability-specific human capital (for example, by entering a less physical occupation, or learning to use adaptations), which should reduce the disadvantage experienced over time. Consistent with this, the impact of disability has been found to be greater among older onset groups across several countries, including the US, the UK, and Australia.