Select Committee on Work and Pensions Written Evidence

Memorandum submitted by Professor Richard Berthoud


  Everyone knows that disabled people are less likely to have a job than non-disabled people. But what is it that distinguishes between disabled people in and out of work?

  This submission is motivated by a contrast: between the intensity of the policy interest in the economic position of disabled people, and the lack of detailed understanding of the relationship between disability and employment. It is based on research which will be published by the DWP next month. The full report is attached, in advance of publication, with the kind permission of the Department. The submission does not comment directly either on incapacity related benefits or on Pathways to Work, except in so far as an understanding of the variations in disabled people's employment prospects contributes to the principles on which policies should be based.

  The new research, and this submission, are based on the Health and Disability Survey (HDS), a detailed survey of disabled (and non-disabled) people undertaken by the then Department of Social Security in 1996-97. The information is not very up to date, but it is the latest available at this level of detail. Other sources suggest that the situation has not changed greatly since then.


  It is common to quote a single overall average for the proportion of disabled people (of working age) who have a job. A first point is that the overall average of 50% used by the DWP for its Public Service Agreement targets (and quoted by the Committee in announcing this enquiry) is highly misleading. That figure is based on the Labour Force Survey definition of disability which is far too loose (ie it includes many people with minor health problems and high employment rates). A more realistic figure for the overall employment rate of disabled people is 30%.

  But the main message is that any estimate of the overall average employment rate of disabled people can be misleading. It encourages analysts and policy makers to think of disability as a single condition, with little variation in prospects between disabled people. Such a monolithic view leads to assumptions about the position of "most" disabled people. And it diverts attention way from disability itself towards other characteristics of disabled people such as their employment history, their educational level or the benefit regime they encounter. These are all important issues, for disabled as well as non-disabled people. But the single most important influence on their employment is their disability.

  The research shows that three sets of disability characteristics are important:

    —  Condition—that is the type of disease or body-part affected. Musculo-skeletal conditions are easily the most common, but mental illnesses have the most adverse effect on employment.

    —  Type of impairment—that is, the normal tasks that people are restricted in performing. Note that in general it is not possible to assume that any particular condition leads to any particular type of impairment. Locomotor impairments are both the most common, and have the worst influence on job prospects.

    —  Severity—that is, how serious is the restriction on performance of normal tasks. Again, severity cannot be inferred directly from the conditions people report. In general, the more severe the impairment, the less likely people are to have a job.

  These three packages of variables make a significant contribution to an explanation of employment probabilities across the working-age population as a whole, and especially to an explanation of variations within the disabled group. In the latter case, variations by disability characteristics are more important than any other sources of variation identified by the survey.

  The personal model of disability would interpret these findings to mean that people with certain conditions and impairments cannot work. The social model would interpret them to mean that employers will not offer them a job. Adherents of the social model often distrust analysis which distinguishes between disabled people according to the nature of their impairments. But it seems likely that employers' willingness to offer a job will vary by disability characteristics. Given that impairment is a necessary precondition for disablement, the social model lacks a framework to account for variations in outcomes between people with different types and severities of impairment.


  The fact that disability characteristics affect variations in employment rates among disabled people does not mean that other characteristics are irrelevant. The analysis has shown that the demographic (especially family) characteristics which so strongly affect everyone's employment prospects affect disabled people too. It has also shown that economic characteristics (especially education) affect disabled and non-disabled people alike. The economic influences measured by the analysis are relatively weak, though it seems likely that they would have seemed stronger if better data about economic characteristics had been available. Indeed, economic influences seem to be more relevant to severely disabled people than to the rest of the population—a severely disabled person with adverse conditions and impairments is much less disadvantaged by his or her impairments if s/he had a good education and lives in a prosperous area, than if s/he had minimal education and lives in a depressed area. This finding discourages too literal an interpretation of the idea of that many individuals are made "incapable" of work simply by their impairments.


  Of course, each disabled person either has a job, or does not. But we can use a "logistic regression equation" to identify groups of people with favourable characteristics and a high probability of employment; and conversely groups with unfavourable characteristics and a low probability.

  The average employment probability of non-disabled people is 76%. (This average is the same as saying that 76% of non-disabled people have a job.) Because disabled people tend to have some unfavourable demographic and economic characteristics (older, less-educated), the average employment probability of disabled people would be only 69%—seven percentage points lower—even if they were not disabled. But the actual average among disabled people, including consideration of their disability characteristics, is only 29%. We could say that the "employment disadvantage" associated with disability—the reduction in their job chances—is 40 percentage points.

Figure A


  The averages for non-disabled and disabled people conceal wide variations in job-prospects.

    —  The thin grey line in Figure A shows that the great majority of non-disabled people have characteristics which give them a high chance of employment—most of them in the 80 or 90% region—though there is a tail of disadvantaged non-disabled people with very poor job prospects.

    —  The dotted black line in Figure A plots the probability of having a job calculated for disabled people if their demographic and economic characteristics are taken into account, but their disability characteristics ignored. This is equivalent to estimating what their chances would be if they were not disabled. Because more of them have disadvantaging characteristics, the number who would have very good prospects is smaller than among the non-disabled group, and the number with poor prospects is larger. But the shapes of the two distributions are not hugely different from each other.

    —  The solid black line in the figure plots the actual distribution of employment probabilities among disabled people, including the effect of their disability characteristics. The shape is completely different. Very few are so mildly disabled as to have good job prospects. Many are so severely disabled as to have minimal job prospects. A substantial proportion of disabled people's chances are so low as to be close to zero. Half of them for example, have a probability of less than 22%.

  Figure A illustrates the width of the gap between disabled people's experience and what they might have expected if disability had not been a source of disadvantage. It is this gap which can be labelled the employment disadvantage associated with disability. The "logistic regression" formula allows us to calculate the disadvantage faced by each individual, and the distribution is plotted in Figure B. The average reduction in employment rates (always comparing disabled people with their own alternative prospects) is 40 percentage points. The range of variation is quite wide, with a small number of disabled people even appearing to enjoy a small gain in their probability, and a small number having their chances virtually wiped out by their disadvantage. It would be appropriate to describe those at the right hand end of the distribution as "those who can work" and those at the left hand end as "those who cannot". But the most striking characteristic of the distribution is the smoothness of the "hill", with its central peak and symmetrical slopes.

Figure B


  Much policy discussion is based on a set of assumptions about the profile of impairments experienced by IB claimants. The Pathways report, for example, said that "the large majority do not have severe conditions". But it is very difficult to reach conclusions about the experience of "most" claimants. The analysis of employment probabilities shows a wide range of employment disadvantages. There is no bulge in the distribution which could be used to argue that "most" disabled people are at the less-severe end of the spectrum; nor a bulge at the opposite end which would suggest that "most" of them are "incapable of work". Still less is there a pair of bulges which could conveniently be used to distinguish "those who can work" from "those who cannot". All we can say is that "most" disabled people face a significant disadvantage; that there are some who are less disadvantaged, and others who are even more disadvantaged.


  A weakness of the disability survey, based on a single set of interviews, was that it provided hardly any information about the process of transition at the entry to disability. Most of the analysis had to be based on an assumption that disability is a source of economic disadvantage. But there are also clear signs that economic disadvantage is a source of disability. The economic characteristics (years of education and local labour demand) which help to predict whether people have a job, also help to predict whether people are disabled, and even how severe their impairments are. Some of the apparent association between disability and losing one's job may really be an association between not having a job and becoming disabled.

  The survey showed that the great majority of disabled people not in work report that they are "permanently disabled" when asked to record their economic position. But the proportion of non-working disabled people who said they were permanently unable to work rose substantially between 1985 and 1996-97. There is a clear sign here of the growth of disability as an economic identity; the acceptance of "I am disabled" as an appropriate economic role by the disabled person, his or her personal community (family and friends), and perhaps the broader public community (including employers, doctors and benefit administrators). (Just as "I am a housewife" was accepted as an economic identity 30 years ago.)

  These "dynamic" questions about the process of becoming disabled and leaving the labour market are clearly an important part of the overall story, even though they are poorly described by the data-set used for this paper. But recent policy discussion has perhaps overemphasised the dynamics of the process, and played down the underlying relationship between impairments and employment probabilities. One of the conclusions of this analysis has been that impairment is an important part of the overall story, too.


  None of this analysis invalidates the Pathways approach, which aims to reduce employment disadvantage through a combination of rehabilitation, labour market engagement and financial incentives, and which is already showing some clear signs of success. The objective might be expressed in terms of countering the growth of disability as a long-term "economic identity" (as discussed above). Perhaps, though, the analysis should encourage government to adopt rather less ambitious targets, at least in the short run. If it was thought that "most" IB claimants had minor impairments, it might have been hoped that "most" of their employment disadvantage could be eliminated, and that the majority of claimants could return to work. Reducing employment disadvantage seems a more realistic objective, which would result in "many" more claimants returning to work.

Professor Richard Berthoud

November 2005

previous page contents next page

House of Commons home page Parliament home page House of Lords home page search page enquiries index

© Parliamentary copyright 2006
Prepared 6 May 2006