Official statistics and the manipulation of conceptual and technical instruments: implications for research on social security
The use of conceptual and technical instruments
The term 'conceptual instruments' describes the categories used in producing a report (Scott, 1990). The term 'unemployment', for example, is the conceptual instrument used to describe and define what 'unemployment' is. Conceptual instruments are important in administrative routines that produce official reports and statistics. They are also important for social research, in so far as the latter uses such data, particularly for secondary data analysis. The term 'technical instruments' refers to the specific methods used to collect information (Scott, 1990). The current technical instrument used to collect data on unemployment, for example, is the unemployed count.
A recent study has revealed the extent of political interference in the compilation and publication of official reports, with particular references to the Social Fund, young people, and black and minority ethnic groups (Craig, 1998). This article is based on my research into social security provision in Britain over the 1980s and '90s, and draws attention to the problems created by the manipulation of conceptual and technical instruments. The main reports used in my research included recurrent, regular and special records on people registered as unemployed, on claimants and recipients of social security benefits, and on social security benefit expenditure, produced by the Department of Employment (DoE), the Department for Education and Employment (DfEE), the Office for National Statistics (ONS) and the Department of Social Security (DSS). All official records, and recurrent ones in particular, are expected to be comprehensive, continuous, and reliable. Thus, the conceptual instruments, which are used to define the categories, should also be continuous. For example, in order to have continuous, comprehensive, and reliable records of unemployment, we need a standard definition of what 'unemployment' is.
Manipulation of conceptual and technical instruments: the case of the unemployment count
During the 1980s the Government changed the definition of unemployment from register unemployment to claimant unemployment (see Table 1). Technical instruments in collecting data for the unemployed also suffered the problems of discontinuation and manipulation. This was because throughout the 1980s and '90s, governments identified the conceptual with the technical instruments: the registration of the unemployed became the definition of unemployment and the means to count unemployment. In order to make things clearer, we have to compare the definition and procedures of unemployment count used by the Department of Social Security (DSS) and the International Labour Office (ILO). The latter defines as unemployed anyone who is over the age of 16 without a paid job but available to start work within two weeks and who has looked for a job at some point of time during the four weeks prior to the interview (DoE, 1993a, p. 312). In order to collect data on the number of unemployed people, the ILO interviews a sample of people. Thus, in this case, we have a distinct separation of conceptual and technical means.
As we can see from Table 1, since 1983 the unemployment rate is expressed as the number of people aged 18 and over claiming unemployment-related benefits (e.g. Unemployment Benefit, Income Support, Supplementary Benefit, Jobseekers Allowance) as a percentage of the total workforce (the total number of people in employment, the self-employed, the unemployed as counted in the relevant year, HM Forces and people in work-related government training programmes). A first result of such manipulation of the definition of unemployment is the discontinuation of the official records: the comparison of the unemployment records prior to and after 1983 are ineffective. In reality, these records measure two different things: registered unemployment and claimant unemployment. In practice, researchers do compare these records, often being unaware of the difference. If we are to compare or use the official rates of unemployment in research (and we have to, since they are the official, and often the only, records), we have to be aware of the fact that these records are both discontinuous and unreliable.
Click HERE to see Table 1: Definition of unemployment
A second result is a technical fall in the rate of unemployment. The exact number of unemployed people is not known: many unemployed people became 'invisible', since they were excluded from the register if they did not claim benefits. Yet, it is clear, claiming unemployment benefits is not the same as being unemployed, and claimants are not the only unemployed people. All systems of measuring unemployment, prior to and after 1983, failed to include specific groups of unemployed people such as those who leave their jobs voluntarily, those who have been dismissed for industrial misconduct, or are considered by the Benefit Agency to have brought unemployment upon themselves; the self-employed (who, although they are not counted as unemployed are included in the total workforce and thus, increase the denominator by which unemployment is counted); those who choose not to register as unemployed; and those who are considered as unavailable for work by the DSS. In addition to these groups, claimant unemployment fails to record the unemployed people who are not eligible, for any reason, for benefits. The DoE's Labour Force Survey (LFS) provides comparisons of the official unemployment rate and the unemployment rate based on the ILO's definition. (However, since the unemployment statistics were moved to the ONS, the Labour Market Trends provides ILO unemployment statistics alongside claimant unemployment statistics). Using this definition, the LFS calculated that in 1994 there were 61,000 unemployed people excluded from the official register in the first quarter, 154,000 in the second, 109,000 in the third, and 103,000 in the last (DoE, 1995).
Click HERE to see Table 2: Changes affecting the unemployment count 1979-1988
The underestimation of unemployment rates has been intensified further by continuous changes in social security regulations and benefit entitlement (see Table 2). The most contradictory result of the identification of unemployment with claimant unemployment is that every time the Government made benefit entitlement tighter for unemployed people, it reduced the official rates of unemployment. That meant that unemployed people were not only without a job, they also had their benefits cut and were taken off the unemployment register. The technical fall of the number of unemployed people was as significant as 550,000 in July 1986 and 668,000 in January 1989. These numbers represent the number of persons in Government work-training schemes excluded from the unemployment count (some indicative figures are included in Table 3). They also include people struck off the register as a result of work-availability tests. This was recognised by a report of the Bank of England:
...although unemployment is falling because there are more jobs, it is also true that much of the decline in the claimant count which has occurred since mid-1986 has been due to a shift in the unemployment/employment relationship resulting from changes in the Government's range of Special Employment Measures - especially the introduction of more rigorous availability for work tests and the rapid growth of the Restart programme (quoted in SSAC, 1991, p. 59).
Click HERE to see Table 3: Number of people in 'Training for Work' and 'Youth Training Schemes'
Even more interestingly, the Government not only excluded people in work-related training schemes, it included them in the number of people in the total workforce. This meant that the Government manipulated the ratio of the claimant unemployed to the total workforce, generating a technical fall in unemployment rates. Similarly, it excluded the under-18s from the unemployment count. Thus, if they were unemployed, they could not claim benefits and they were not included in the count. Once in employment though, they were included in the total workforce, affecting the ratio between claimant unemployment and total workforce. The DoE (1990:S18) estimated that since 1986, an annual average of 90,000 under-18s were excluded from the register due to their inability to claim benefits. The exact number of people excluded from the unemployment register is difficult to estimate. The Social Security Advisory Committee has provided an estimation (see Table 2). Even this account, however, is not complete. It has left out, for example, students. From November 1986 to September 1990, students could claim unemployment-related benefits during their summer holidays (prior to that, they could claim during any holiday, given they were actively seeking work), but they were not included in the unemployment count. That meant that they were recognised as unemployed, since they were eligible for unemployment benefits, but only unofficially. To resolve this contradiction, since 1990, students have not been allowed to claim unemployment benefits at all.
The third, and probably the most important result of conceptual and technical instrument manipulation, is the exclusion of a significant number of people from claiming benefits: hundreds of thousands of people were made ineligible. These were the most deprived in a system which, obsessed with labour supply and wage labour dependency, had refused them both a wage as a result of unemployment and the alternative of social security benefits as a result of managing the accessibility of benefits. This penalisation of the unemployed is not incidental. Social security has not only reproduced labour dependency by managing the availability, accessibility and levels of benefits, but also by developing a penal character applied to those not participating in the labour process. The involuntary nature of unemployment is residual to this function of social security and had become even more so under the Conservative Governments. It seems that the victimisation features of social security are intensified at times of greatest poverty and need (Novak, 1997).
Discontinuation of benefit categories
Similarly, extensive changes in the terms of eligibility and/or the names of benefits were made during the 1980s. These changes have been extensive and have undermined the continuity and, in some cases, the validity of recurrent records and of other reports. For example, Single Payments were replaced by the Social Fund and Unemployment Benefit (UB) and Income Support (IS) for the unemployed were replaced by the Jobseekers Allowance (JSA). When these changes were accompanied by changes in the operation of the benefits, as was the case in both these examples, the comparison between the two different benefits is impossible, thus resulting in a complete discontinuation of statistical tables. Moreover, it is almost impossible to research the impact of the changes on the incomes of a standard group of claimants over a long period of time. The recipients of Unemployment Benefit by category, for example, have changed a number of times making it impossible to keep track of which group has benefited or has lost out due to the introduction of new regulations. Other categories are not directly comparable to previous ones (as in the case of unemployed people and the creation of new categories of recipients after the introduction of the under-25 and over-25 groups of recipients).
More importantly, changes in conceptual instruments have also resulted in considerable changes in registration. Thus, registered unemployment may drop as a result of excluding groups from eligibility rather than an increase in employment rates; the number of people in poverty may fall as a result of a reduction in the number of individuals who are eligible for state benefits rather that an increase of mean incomes; and exceptional needs may seem less acute as a result of the abolition of Single Payments rather than an increase in provision and security. In brief, manipulation of the conceptual instruments used to define categories may result in distortion of statistical data and information. Similarly, a further, and very important, problem of manipulation of conceptual instruments is the changes of categories for Supplementary Benefit (SB) and Income Support (IS) recipients after the introduction of the latter. Until the introduction of IS, there were three main categories of recipients: single non-householders, single householders and married couples. Each category could claim either the ordinary rate, or the higher rate for long-term unemployment. After the introduction of IS, the categories changed altogether to single people under 25 years of age, or lone parents under 18; single people over 25 and lone parents over 18; couples both under 18; and couples with one or both partners over 18. The problem with these changes was that comparative research between groups of recipients before and after 1988 is made extremely difficult.
Overcoming the manipulation
There are a few ways to overcome such problems (although some problems can not be overcome at all). The first is the choice of research methodology. A qualitative analysis of social relations has the advantage of concentrating on the driving forces behind the changes in social security, whereas quantitative changes do not define the analysis itself. For example, researching the operation of Unemployment Benefit can underline the importance of policies that focus on forcing people off unemployment registers rather than focusing on the number of people registered as unemployed. That is not to say that the impact of the changes on the individual claimant or the fluctuation of the rate of unemployment are not important. On the contrary, it is essential to highlight the deprivation of claimants as a result of government policies. It is rather to say that the focus of the research should be on the quality of the provision rather than, necessarily, on the numbers of eligible people.
A further solution to the problems described above is the choice of the sources themselves. Official statistics may be only one source of information. At the same time, other alternative information may be available: for example, statistics provided by the ILO which uses its own standard definition of unemployment and presents continuation in its statistics. Similarly, the DSS historical supplements provide some adjusted data on the number of unemployment claimants (though they do not provide such adjusted statistics for other groups of claimants) (DoE, 1992; DoE, 1994e). The DSS and other organisations also provide some information on the initial impact of changes in social security, as, for example, in calculating the number of people excluded by the introduction of new benefits or regulations from recipient. However, the exact number of claimants affected by the introduction of new benefits, the abolition of others, or by changes in regulations (for example, some recipients are excluded altogether, some lose in money terms) are not often available. Nevertheless, this may be less important in cases where the aim of research is to provide an understanding of the way social security and social relations interact. That is, where it pays primary attention to social relations and to the qualitative analysis of this interaction, and secondary attention to case studies or the quantitative impact of the changes.
Craig, G. (1998), 'The Privatisation of Human Misery, Critical Social Policy, 18(1), pp. 51-76.
Department for Employment and Education (1995b), Employment Gazette, October.
Department of Employment (1988a), Employment Gazette, January.
Department of Employment (1988b), Employment Gazette, March.
Department of Employment (1988c), Employment Gazette, April.
Department of Employment (1988d), Employment Gazette, December.
Department of Employment (1990), Employment Gazette, December.
Department of Employment (1992), Historical Supplement No. 3 Employment Gazette, June.
Department of Employment (1993a), Employment Gazette, July.
Department of Employment (1993b), Employment Gazette, December.
Department of Employment (1994e), Historical Supplement No. 4 Employment Gazette, October.
Department of Employment (1995a), Labour Force Survey Quarterly Bulletin, December.
Novak, T. (1997), 'Hounding Delinquents', Critical Social Policy, 17(1), pp. 99-109.
Scott, J. (1990), A Matter of Record: Documentary Sources in Social Research, Polity Press: Cambridge.
Social Security Advisory Committee (1991), Social Security Provision for the Unemployed. A Report to the SSAC, HMSO: London.