Report of the Meeting of Experts on Labour Statistics
(Geneva, 14-23 October 1997)
1. At its 267th Session (November 1996), the Governing Body of the International Labour Office decided to convene a Meeting of Experts on Labour Statistics. The Meeting was held in Geneva from 14 to 23 October 1997.
Agenda of the Meeting
2. The agenda of the Meeting was the following:
I. Underemployment: concept and measurement.
II. Income from employment: concept and measurement.
3. The first item on the agenda was discussed from 14 to 17 October and the second from 20 to 23 October 1997.
Participants
4. Twenty-six experts were invited to the Meeting, 14 following consultation with governments, six nominated by the Employers' group and six nominated with five who attended by the Workers' group of the Governing Body. Nine observers and five representatives of intergovernmental organizations also attended the Meeting. The list of participants is annexed to the report.
Opening address
5. The Meeting was opened by Ms. Chinery-Hesse, Deputy Director-General of the ILO. She recalled that the role of the Meeting was crucial to help the Office to prepare a report on the revision of international guidelines on concepts and measurement of underemployment and new guidelines on concept and measurement of income from employment. This report would be placed on the agenda for submission to the Sixteenth International Conference of Labour Statisticians (ICLS), subject to the recommendation and approval of the Governing Body of the ILO.
6. Mr. Mehran, Director of the Bureau of Statistics, referred to the absence of formal rules of procedures, which was meant to facilitate a free exchange of views among the experts present there. Delegates were participating at the Meeting as individual experts, not necessarily representing views of their respective governments or organizations. He reminded participants that the Meeting was expected to arrive, to the extent practicable, at an agreed statement of its conclusions that would help the Governing Body to take a decision concerning future action on the subjects.
Report I -- Underemployment
(14-17 October 1997)
Election of the Chairman
7. Mr. Ian Macredie, the expert nominated by the Government of Canada, was elected Chairman of the Meeting for agenda item I. He congratulated the Meeting for its tripartite representation and the commitment of all parties to accomplish the important task before it. The challenge was to add clarity to standards which were in existence since 1957 to enable more countries to apply them. He welcomed this opportunity for discussion of a subject on which the interests of both industrialized and developing countries clearly converged.
Presentation of the working document
8. In presenting Report I to the Meeting, attention was drawn to the need for data on underemployment, the status of current statistical standards and why they are not widely applied today. Underemployment reflects a situation in which full employment, as understood by ILO Convention No. 122, is not achieved. The purpose of revising the current standards was to encourage wider measurement and to promote international comparability. One important feature of the proposed definitions was consistent integration into the labour force framework, which sets the rules for classifying persons in employment, unemployment and economic inactivity and is recognized and applied worldwide. Another important feature was the fact that many countries already were measuring underemployment using similar definitions.
9. Two definitions were being proposed, a revised definition of visible underemployment, based on more precise criteria than in current standards, and a definition of a broader concept of underemployment, which encompasses visible underemployment. Both definitions would require workers to express willingness and ability to change their work situation, a central requirement in the labour force framework.
Terminology
10. The Meeting noted that several of the terms used in the proposals conveyed a different content than was given in the definitions of these terms. For example "taking additional employment" conveyed the idea of obtaining another employment in addition to the current employment. However, the definition also included the replacement of the current employment with another employment with increased hours or an increase in the hours of work of the current employment. Similarly "taking alternative employment" conveyed the idea of replacing the current job for another one, but the definition also included obtaining an additional job or modifying the current job. The Office should make efforts to find more appropriate terms. For example, in French "emploi suppletif" would seem preferable to "emploi alternatif".
11. In contrast, the term "inadequate employment" conveyed a broader notion than what the actual definition proposed, including aspects relating to the conditions of employment and social welfare which go beyond the underutilization of labour resources. It was for this reason that the Meeting agreed that what the report before it termed "inadequate employment" be renamed "underemployment" and leave the term "inadequate employment" to indicate the broader notion, including conditions of employment and social welfare.
General discussion
12. During the discussions, a number of issues were raised. There was broad agreement concerning the need to arrive at a measure of underemployment and of visible underemployment, even if they are not perfect. These measures have long been relevant in developing countries and are of increasing relevance in industrialized countries. They should be suitable to address issues which are useful and valid for practical policy decisions, and should relate to the underutilization of the labour supply. Other issues concerning social welfare are related to these and could also be addressed. The absence of comprehensive measures of underemployment might provoke policy-makers to fill the data gap with less appropriate indicators. One consequence of this might be to put undue pressure to revise the current definitions of employment and unemployment. It was mentioned that good data on underemployment could help to explain the seemingly contradictory phenomenon of labour migration taking place from countries with low unemployment rates to countries with higher rates. In addition, underemployment due to lack of skill utilization was seen to be a waste of human resources, particularly relevant to women.
13. Another issue which received strong support was the integration of the underemployment concept within the labour force framework because this will align it with the internationally accepted concepts of employment and unemployment. One consequence of this alignment is its basis in the workers' perception of their own work situation and not in the employers' perception of the utilization of labour resources. Another consequence is measurement in reference to a short period. The labour force framework being best applied through labour force surveys, a third consequence is that these are the natural source through which underemployment can be measured, thus establishing a common measurement methodology.
The conceptual framework
14. It was generally agreed that there are different dimensions in underemployment, related to hours worked, earnings per hour and skill utilization, etc. Other dimensions which could be considered as affecting productivity of work mentioned were working conditions and job satisfaction. In this context, it was useful to consider visible underemployment as an essential dimension of underemployment. Disaggregating the measure of underemployment by its various dimensions would be useful because it would allow users flexibility of data for different policy objectives.
15. Overemployment, relating to situations where workers work long hours each day in order to make a living, can be seen as a form of underemployment on the dimensions of income and skills, especially in developing countries. Of interest would be to identify those multiple job-holders who have taken secondary jobs because of insufficient hours in the main job.
16. It was also agreed that underemployment was to be defined in relation to each and every job of the worker. Persons would be visibly underemployed if they were willing and available to work beyond the hours usually worked in all jobs. In contrast, persons would be in other forms of underemployment with respect to any of the jobs (main or other).
17. Most opinions converged on the idea that underemployment should be measured in relation to an alternative employment. This is the approach used to define unemployment in the labour force framework. In contrast, the use of norms as a definitional criteria received mixed reactions. A number of experts considered norms as a hindrance to international comparability. Within a country, they also present problems for the selection of an acceptable norm: when set in relation to national legislation (e.g. normal hours of work, minimum wages, etc.), any change in legislation or its application will generate a change in the number of underemployed persons which does not necessarily reflect changes in the underlying reality. Additionally there are countries where legal norms may not be significant, or they may not be relevant for certain groups of persons. When set as a value of central tendency it may bias the results if the economy as a whole is in a situation of under or overemployment. However, other experts viewed the use of norms as fundamental for policy decisions which are always linked to national labour standards. For these reasons, it was suggested that if norms are used, they should not be set at the data collection stage which would limit the scope of the information obtained, but applied at the tabulation stage.
18. In order to reflect gender issues more clearly, the Meeting agreed on the need to disaggregate underemployment data by sex and age group. Underemployment is thought to affect women in particular. In this respect the proposal of the secretariat also recommended to disaggregate data by presence and age of children in the household.
19. It was recognized that certain groups of workers might be difficult to capture and identify as underemployed through this labour force framework: seasonal workers in rural areas who, when they work, work very long hours and are otherwise inactive, contributing family workers, child workers and other workers with low wages who have limited possibilities to obtain alternative employment.
20. Concern was expressed that underemployment estimates as proposed would behave in contrast to the economic cycle in the sense that during economic recovery more people change jobs, and therefore that more persons would be identified as underemployed during those periods. For the same reasons, the relationship between underemployment and labour market flexibility was expected to be positive. It was observed, however, that countries where a measure which is similar to underemployment as proposed is carried out, it has been seen to follow a similar pattern over time as their unemployment measure.
21. There was a suggestion that if the main objective was international comparability, the adoption of precise international standards on underemployment could be premature. It might be preferable to give emphasis first to the adoption of a sound conceptual framework which would then be tested in a number of countries. The experience gained would help decide whether a common international definition could be developed. Although the need for further experience in the application of the proposed criteria, possibly in the form of a structured programme sponsored by the ILO, was strongly supported, it was also felt that adopting an initial international definition of underemployment would be beneficial for this purpose. The experience of countries which already measure underemployment using similar criteria as the proposed definition would be essential.
Visible underemployment
22. The proposal before the Meeting on visible underemployment was based on three criteria: workers should have worked less than the normal duration of work during the reference period and should be willing and available to take additional employment (i.e. to work more hours in an additional job, in the current job, or in another job which replaces the current job). Normal duration of work was further defined as the modal value of workers' usual hours. Workers' willingness could be operationalized by requiring workers to express a preference or desire for additional employment or to be actively seeking additional employment. This last criterion aligned the definition with the current labour force framework while maintaining continuity with the past.
23. There was agreement that it was essential for the definition of visible underemployment to be anchored on workers' willingness and availability for additional employment. Willingness in particular was felt to be statistically more reliable than the involuntariness criterion used in the current international definition, which it substitutes, and which made the proposed definition of visible underemployment more practical than the current one. In particular, that definition will be able to identify persons declaring voluntary reasons for working few hours who nevertheless are able to express their willingness to take additional employment, for example, persons who declare their reasons for working few hours as family constraints who declare willingness to work additional hours. They would be classified as visibly underemployed, according to the proposed criterion but not according to the existing one.
24. It was generally accepted that applying the first criterion, i.e. that workers should have worked below a certain number of hours during the reference period, should be left to countries to determine in line with their policy needs but the resulting figures should not be used for international reporting. The chosen limit should be applied at the tabulation or analytical stage, rather than be embedded in the measurement process. The problems of determining normal duration of work for workers with special time arrangements (workers on annualized hour schedules, shift workers, teaching professionals whose hours are well below normal duration of work) would thus be minimized. There was controversy on whether the norms should be determined on the basis of legal limits or in terms of values of central tendency based on survey results. Legal limits may be different for different workers, and different than values of central tendency. Furthermore, values of central tendency like the one proposed may not be representative of normal duration in cases where part-time employment is widespread. In both cases, adoption of a uniform norm on hours of work may not be meaningful, even within a country.
25. "Seeking additional employment" as defined in the proposal, was considered to be the most suitable measurement criterion to determine visible underemployment because it is objective and operationally feasible. At the same time it was recognized that in order to understand labour market inadequacies more fully it would be necessary for policy-makers to also obtain information based on workers' expression of desire or preference for additional employment. Longitudinal studies on the behaviour of unemployed and of inactive persons who have not sought work but want to work have shown that both groups have similar propensity to enter employment. The analogy could be made with those persons who did not seek additional work, although they wanted it. If both criteria (seeking and wanting) are to be measured, it is important that international guidelines recommend the order in which questions relating to them are raised in a household survey, because the effect of their position may not be trivial. It is advisable that such guidelines be based on field testing. Experience where both a soft criterion such as wanting additional work and a strict criterion such as seeking additional employment are applied has shown that a soft criterion is also more sensitive to changes in the economic and labour arenas than the strict criterion of actively seeking additional work. However, the soft criterion is also more sensitive to changes in question formulation. In all cases the seeking additional employment criterion may need to be relaxed in situations of largely unorganized or limited markets, restricted labour mobility, or compulsorily reduced hours of work due to shortfalls in the demand for labour or products, or family constraints.
26. Regarding availability for additional employment, different reference periods to determine the readiness of workers to take up additional employment were proposed. The proposals ranged from the same reference period as is used to define the availability criterion in the unemployment definition, to a longer period of one month in order to include workers who need to quit one job to start another.
27. The usefulness of distinguishing between "involuntary part-time workers" and "economic short-time workers" proposed in Report I was recognized. However, the group definitions as proposed did not seem to reflect the groups as they were being defined by countries in the European Union, that is, on the basis of workers' self-assessment of their working schedules (part-time and full-time), and not on the workers' usual hours in relation to a cut-off point. One element of the problem is that, although there is a definition of part-time work in ILO Convention No. 175, it does not lend itself to statistical measurement. It was also mentioned that economic short-time workers should not be required to seek additional employment because their short hours are imposed on them and they are often compensated for the lost wages. In this respect, the proposed definition allows for the seeking additional work criterion to be relaxed under this situation, and this could be made compulsory if consensus could be reached.
28. The issue was raised whether the distinction between these two groups should be based on whether visible underemployment was a person's usual or temporary status rather than on the basis of part-time work. This way, the distinction would be meaningful to all countries, not only to those where part-time work is widespread. This distinction could be done as is being proposed, on the basis of usual hours. It could also be done by inquiring about the duration of visible underemployment.
29. The volume of visible underemployment, defined as the additional hours that persons are willing and able to work, will be closely related to the number of visibly underemployed persons. It is a useful indicator for policy decisions and public debate, especially given the limitations of head counts. However, to be useful it would need to be compared with other volume measures.
30. Longitudinal analysis would allow the monitoring of the behaviour of visibly underemployed persons. Repeated measurement was seen as a means to provide better insight into the problems of underemployment, because it would allow the representation of all important cycles during a year. However, full-time casual workers who are willing and able to work additional hours over a year will not be identified as visibly underemployed any time during the year, even though repeated measurements are made, if each one is using a short reference period. Longer reference periods would need to be considered to capture these and other situations, such as to calculate annual volume of visible underemployment.
31. Multiple job-holding represents a problem for the collection and presentation of data on visible underemployment by industry, occupation and status in employment. It imposes the choice of a reference job amongst the various jobs, which entails defining what a main job is.
32. The visible underemployment rate should be calculated as a percentage of the labour force to maintain consistency with other labour force indicators, such as the unemployment rate.
Underemployment
33. The proposal before the Meeting on underemployment was also based on three criteria: workers should be willing to take alternative employment (i.e. an additional job, a replacement of the current job by another one or the modification of the current job), for reasons linked to an increase in the duration or productivity of work. They should also be able to take alternative employment. Workers' willingness to take alternative employment could be operationalized by requiring workers to express a preference or desire for alternative employment or to be actively seeking such alternative employment. Using this last criterion would align the definition with the current labour force framework. Three reasons had been identified as relating to an increase in the duration or productivity of work. These relate to hours of work, income (from employment) per hour and utilization of skills, but other reasons could be recommended. Ability to take alternative employment could be operationalized in two ways: by requiring workers to be available to take alternative employment or to be ready to take alternative employment without acquiring new skills.
34. The Meeting found that it was important to measure the wider concept of underemployment, noting that visible underemployment was less relevant than other forms of underemployment in the developing countries. The issue was raised whether underemployment data should be collected with the same or lower frequency as data on employment and unemployment.
35. The same arguments were advanced regarding the need to maintain consistency with unemployment as for visible underemployment. Therefore "seeking alternative employment" was considered to be the primary classification variable. Similarly the need to identify workers who did not seek alternative employment but who expressed the desire or preference for it, was essential.
36. There was general agreement that determining a list of qualifying reasons for wanting or seeking alternative employment was fundamental in delimiting the boundaries of underemployment. Hours, hourly income and skill utilization were generally accepted as valid reasons, if low hourly income could be seen as resulting from underutilization of human resources.
37. Other reasons such as job satisfaction and working conditions were more controversial as they did not directly relate to the underutilization of labour resources. However, they are related to social conditions which have become current issues. The problem of the validity of hourly income as indicative of underemployment was questioned, in cases where skill utilization was not combined with it. Income per hour and utilization of skills should be seen in combination so that low income per hour could be identified as resulting from underutilization of skills. An example where hourly income and skill were not related, however, referred to workers willing to move from the informal to the formal sector or from subsistence agriculture to wage employment in order to increase their hourly earnings without using their skills more fully. Valid reasons could be best determined through field testing of the validity of replies to questions on reasons for seeking or wanting an alternative employment.
38. The usefulness of disaggregating the measure of underemployment by its various dimensions was reiterated. It was recognized that the dimensions would not be additive as persons may be underemployed in terms of more than one dimension at the same time. The identification of the various dimensions would require the possibility of multiple responses to the question on reasons for seeking or wanting alternative employment. Reservations were made, however, about separating the hourly income and skill dimension since they affect each other intrinsically, as well as concerning the reliability of the replies when proxy respondents were used. The effect of proxy responses could also be best determined through field testing of replies to questions on reasons for seeking or wanting an alternative employment. Another way to identify the different dimensions of underemployment was to focus on the main reason for seeking or wanting to take alternative employment, or to rank the reasons. However, both these solutions would lead to loss of information as well as international comparability.
39. Multiple job-holding affects the measurement of underemployment. While underemployment due to hours would need to be determined with respect to all jobs seen together, the other dimensions of underemployment would need to be determined with respect to each and every job.
40. The requirement that workers, who seek or want alternative employment to increase hourly income or to improve skill utilization, should be ready to take up such employment without acquiring new skills was thought by some to be too restrictive. Skills constantly change and any new job may require upgraded skills. One way to encompass this need while simplifying the criterion would be to require that workers be ready to take alternative employment within a specified reference period. This criterion would permit the upgrading of skills but not the acquisition of new skills.
41. Results from the experience of one country measuring underemployment in a way similar to what was proposed, were presented to the Meeting. They showed that the variation of underemployment rates over time followed a pattern similar to that of unemployment rates; that particularly in rural areas underemployment was more significant than unemployment; that more workers were affected by other forms of underemployment than by visible underemployment; and that underemployment rates were higher for women than for men.
42. Regarding the identification of the employed population to be excluded from underemployment through the use of thresholds, such as workers in certain managerial or professional occupations or workers who worked less than a certain number of hours, it was questioned whether the cost of determining them was worth the small number of observations targeted. It was noted that the proposed definition considered these exclusions optional.
43. The issue was raised on how to classify workers who are willing and able to take alternative employment but whose skills are no longer in demand or are obsolete or who refuse or cannot move to areas where alternative employment exists. It was noted that they should be classified as underemployed in the same way as persons who seek work are classified as unemployed regardless of whether they are employable or not. It was observed, however, that in some countries these persons would be classified as inactive when measured using administrative records because of the loss of their unemployment compensation rights.
44. Information on the work history of persons would help understand why persons fall into underemployment. Data on the skill level of previous jobs and on household characteristics would also provide insights into the problem of underemployment.
Other matters
45. The issue was raised whether the resolution would provide guidelines on how to combine the measurement of visible underemployment and underemployment and on a set of question formulations to include in surveys. Such guidelines could be prepared on the basis of the experience gained after testing the measurements.
46. While it was clear that visible underemployment was a subset of underemployment, in order to have consistency with employment and unemployment statistics, and in order not to reduce the quality of the information obtained, recommended practice was to start by measuring visible underemployment separately from other forms of underemployment using a definition specific to it, followed by the measurement of other forms of underemployment.
47. It was felt that the structure of the proposed resolution should reflect that the overall concept was underemployment and that visible underemployment was one essential component of it. This meant that the proposed resolution and report to the 16th ICLS should begin with the operational definition of underemployment, and then define visible underemployment and "other forms" of underemployment which could be referred to at least three dimensions, i.e. hours, hourly income from employment and use of skills. The proposed resolution could mention the possibility to measure certain aspects of inadequate employment, important for social policy, which are beyond underemployment, such as inadequate employment due to lack of job security, adverse working conditions, excessive hours of work, etc. These could be incorporated within the measurement methodology if desired. It was stated that in many developing countries inadequate employment is more relevant to their labour markets than visible underemployment.
Recommendations
48. The Meeting of Experts recommends to the Governing Body of the ILO that the topic of underemployment be included on the agenda of the Sixteenth International Conference of Labour Statisticians planned for 6-15 October 1998. The report that will be prepared by the secretariat for submission to the Conference should be focused on the concept and measurement of underemployment, providing a clear draft definition of "visible" underemployment and extensive elaborations on the definition and measurement of other forms of underemployment including at least the dimensions of income from employment and use of skills. Even though the concept of "inadequate employment" contains elements which go beyond underemployment, it is still an important concept which addresses important labour market and social phenomena. These phenomena require further elaboration at the conceptual level and refinements through experimental data collection in countries at different levels of development. Furthermore, the draft resolution which will accompany the report to the 16th ICLS should be revised in line with the discussion of the Meeting as summarized in the present report.
49. The Meeting of Experts also recommends that:
- the terminology be adopted so that the term "underemployment" when used alone encompasses both visible underemployment and all other forms of underemployment;
- underemployment be integrated into the labour force framework, which sets the rules for classifying persons in employment, unemployment and inactivity;
- underemployment be defined mainly in relation to alternative or additional employment; definitions be based on both the workers' willingness and their seeking activities; that norms be applied for policy uses only at the tabulation or analysis stage;
- measurements of underemployment distinguish between its different dimensions, including hours of work, hourly income from employment and use of skills;
- qualifying reasons for wanting or seeking alternative employment be determined to clearly fix the boundaries of underemployment;
- a structured programme sponsored by the ILO to test the definition and approach to the measurement of underemployment in a number of countries ranging from developing, transitional, industrialized, be launched; that the ILO consider supporting testing which goes beyond underemployment to other aspects of inadequate employment.
Report II -- Income from employment
(20-23 October 1997)
50. The second item on the agenda was introduced by Mr. F. Mehran, Director of the Bureau of Statistics. He welcomed the participants to the session on income from employment and reminded them of the overall objectives of the Meeting.
Election of the Chairman
51. Mr. J.B. Coker, the expert nominated by the Government of Nigeria, was elected Chairman for this part of the Meeting. Mr. Coker expressed his satisfaction with the challenging and productive discussions which had taken place on the topic of underemployment. He reminded the participants that the convening of this Meeting was a result of the 15th ICLS's recommendation and that the Meeting was expected to arrive at an agreed statement which would help the Governing Body to take a decision concerning future action by the Bureau of Statistics on the measurement of income from employment.
Presentation of the working document
52. In the presentation of Report II to the Meeting, attention was drawn to the need for guidelines on statistics on income from employment. Two existing international standards were related to the measurement of income: one concerned the measurement of earnings, and the other, that of household income. Neither of them provided adequate guidelines on the statistical definition and measurement of the full income accruing to workers by virtue of their employment situation. There were at present no guidelines on the measurement of income from self-employment for the purposes of labour statistics. In practice, statistics of income from employment were rare; they were not yet part of the regular wage statistics programmes of developed or developing countries. Where they had been compiled, they followed a variety of concepts and methodologies and very often had not been sufficiently exploited. There was a need to develop new international standards encompassing all types of income earned through both paid and self-employment activities.
53. Two major objectives had been identified for the measurement of income from employment: the first was to identify economic activities which were more or less remunerative than others; the second objective was to identify those groups of workers who were unable to derive a certain level of well-being from their involvement in economic activities. Because the conceptual aspects of income from paid and self-employment differed, it was proposed that these two topics be examined separately and in parallel.
General discussion
54. The participants felt there was a need for the measurement of income from employment with a view to reflecting structural changes and the increased flexibility of the labour market. Data on income from employment would be particularly useful in analysing such issues as the changes in employment patterns, the growing incidence of precarious employment and the changes in remuneration practices which have taken place in industrialized, transition and developing countries. Information on income from employment should also help improve the measurement of income in the agricultural sector and in the informal sector.
55. The Worker experts identified three key elements that required attention: (i) at the conceptual level, there was a need to distinguish between the productivity and the welfare aspects of the measurement, in order to identify which of the components of income should form part of the concept of income from employment; (ii) at the measurement level, there was a need to determine the monetary value of non-wage benefits within the employer-employee relationship; (iii) at the data collection level, there was a need to improve inquiries and surveys on the components of non-wage benefits.
56. The Employer experts at first questioned the need for new standards on income from employment. The development of new standards was a complex undertaking, while the requirements of existing standards relating to the measurement of individual earnings and labour costs were still difficult to fulfil.
57. It was recognized that a clear definition of income from employment was useful in analysing the labour force situation. There was a need to link income from employment to information related to employment, underemployment and unemployment, as well as productivity.
58. It was noted that the assessment of economic well-being involved not only the measurement of the employment and income from employment situation of individuals on their own, but also that of other types of income, and their evaluation within the context of the household. However, data on income from employment could be used as a significant input in the assessment of the concept of consumption capacity. It was also felt that data on income from employment could complement aggregated national accounts by providing information on the distribution of income in the population covered.
59. There was general agreement that the concept of income from employment referred to the amounts which accrue to employed persons as a result of their involvement in paid or self-employment activities. Concern was expressed as to whether income from employment was primarily meant to cover the receipts that are "perceived as income" by the recipients. After extensive discussion, the participants agreed to remove from the provisional definition the phrase "which are perceived as income by the recipients". This phrase introduced an element of subjectivity in the identification and measurement of income components, and was not applicable to the measurement of income from self-employment, as defined in this report.
60. The Worker experts were of the opinion that income from paid employment should be based on the concept of current income and should refer to the employee's current status in employment.
61. An important point mentioned was that the issue of the reference period should be brought upfront at the conceptual stage. The reference period was not just a measurement issue. The choice of the reference period could differ depending on whether the emphasis was measuring income from employment from the productivity aspect (short reference period) or the welfare aspect (long reference period). There was also the need to relate employment statistics to income from employment statistics, which required the reference periods to be harmonized.
62. It was generally accepted that accurate data on income from employment could not be collected from a single source or survey. Neither establishment surveys nor household surveys alone could provide adequate information on all the various components of income from employment. In the measurement of income from employment a number of additional difficulties should not be underestimated, including the evaluation of benefits in kind, the reference period used for the measurement, the underreporting of income in household surveys and their use of proxy respondents.
63. The Meeting took note of the statement presented by a representative of the International Federation of University Women. She raised the issue of gender measurement in national and international statistics, with a view to better revealing the contribution of women to economic development through the valuation of unpaid work.
Income from paid employment
64. There was general agreement on the proposed definition of income from paid employment, although differing opinions were expressed regarding the components themselves.
65. The participants agreed that the statistics of income from paid employment should cover all categories of persons classified in paid employment according to the latest version of the International Classification of Status in Employment (ICSE).
66. As regards the components of income from paid employment, there was general acceptance that all the components of remuneration in cash should form part of the concept. In addition, it was felt that remuneration in kind should include all forms of payments concerned, even if certain items were difficult to evaluate. The fact that some benefits in kind were provided to employees in a collective manner was not a sufficient reason for exclusion from the concept and measurement of income from paid employment.
67. The expert from the General Confederation of Trade Unions (GCTU), representing the workers from the Commonwealth of Independent States, identified a form of compensation in kind which raised conceptual and measurement problems. This form of compensation consisted of "imposed" payments in kind in lieu of cash remuneration. Some of the participants felt that, in view of ILO Convention No. 95, Protection of Wages Convention, 1949, this type of payment was not to be considered as income from paid employment. Some others felt that, if it was to be considered as income from paid employment, then "imposed" compensation in kind should be given a zero value.
68. The Employer experts stated that the ideal solution would be to collect the value of non-wage benefits from the employees themselves; however, this was perceived to be difficult. Other participants, including the Worker experts, felt that the capacity of employees to attach a value to non-monetary benefits should not be underestimated. In normal collective bargaining processes, monetary values were often assigned to non-wage benefits. However, it was pointed out that employees may find it difficult to clarify the difference between an employee benefit in kind and items provided by the employer within the context of the work environment. It was also noted that tax data often provided a source of valuation of non-wage benefits.
69. As regards social security benefits, the question was raised as to the criteria to be used for determining the inclusion or exclusion of certain benefits from the definition. It was made clear by the Meeting that social security benefits could only be included if they were related to the status in employment, and that social assistance not related to employment should be excluded. However, it was pointed out that it was not always easy to determine which social security benefits were employment-related and which benefits were not.
70. The Employer experts proposed that a distinction be made between social security benefits that were of a permanent nature (e.g. pensions) and social security benefits that were temporary and contingent upon paid employment status (e.g. compensation for occupational injury, maternity and temporary unemployment). While the former should be excluded from the concept of income from paid employment, the latter should be included. It was further suggested that information on those social security benefits that were excluded from the concept of income from paid employment, be collected as part of other income components.
71. The Worker experts and a few other participants proposed that short-term benefits paid through social security schemes or the State (such as unemployment benefit, maternity leave benefit, occupational injury allowance) should be excluded from income from paid employment. These benefits were not determined within the framework of the employer-employee relationship. They reflected the social dimension of government policies and were not viewed as a return to productivity. Only benefits that were fully financed by the employers and provided by schemes other than social security and state schemes should be included in income from paid employment.
72. All other participants were of the opinion that social security benefits, regardless of the source of payment, should be included in the concept of income from paid employment, as proposed in the original definition. Similarly, it was considered that severance, termination and redundancy pay could be assimilated to unemployment benefits and therefore included in income from paid employment.
73. There was agreement that compensation for medical and health expenditure clearly represented a significant benefit to employees and their family members covered by such schemes and should be included in income from paid employment.
74. Extensive discussions took place on the measurement of profit-sharing bonuses and profit-related pay. The participants agreed that profit-related bonuses should be included in the definition of income from paid employment. As regards the distribution of shares by employers, it was generally agreed that employment-related income was represented by the value of the share at the time of distribution to the employee provided the employee is able to sell the share. It was further agreed that the value of the share should be assessed at the current market rate at the time of transfer to the employee net of the amount paid, if any, by the employee. It was then considered that neither subsequent property income or capital gains from the shares, nor losses, should be included in the measurement of income from paid employment.
75. It was considered that, in theory, income from paid employment should be recorded net of employees' occupational expenses. This would maintain consistency between the measurement of income from paid employment and that of income from self-employment. However, in the measurement of employees' occupational expenses significant concerns were raised about the lack of an objective basis for evaluating such expenses and the negligible impact these expenses have on employees' capacity to consume.
76. Discussions ensued on the types of occupational expenses that could easily be quantified, such as day care. It was considered that there was little value added in measuring a single component of these expenses, which, in this case was of little relevance to developing countries. It was therefore agreed that income from paid employment should be measured gross, without any deductions for employees' occupational expenditure.
77. There was some debate about the best way to measure income from paid employment, i.e. whether gross or net of employees' own contributions to pension and social security funds. In theory, social security benefits received should be recorded on a net basis after deduction of the employee's contributions to the relevant schemes. However, it was generally more practical to measure gross income from paid employment before employees' contributions were deducted, although in some situations it would only be possible to measure net income from paid employment after deduction of employees' contributions. The Worker experts reiterated their position on the statistical treatment of social security benefits and stated that if these benefits were excluded from the components of income from paid employment, the issue of net and gross measurement would not arise.
Income from self-employment
78. In introducing this topic, the secretariat drew the Meeting's attention to the fact that the focus of the measurement of income from self-employment was placed on the income of persons who were the sole or joint owners of unincorporated enterprises. In the System of National Accounts, the income arising from the productive activities of an unincorporated enterprise was described as mixed income.
79. There was general agreement that there was a need for statistics of income from self-employment in industrialized, transition and developing countries. It was also recognized that the measurement of income from self-employment was a daunting task. The self-employed constituted a heterogeneous group which included owners of small-scale and larger enterprises run with either little or significant capital, with or without the help of employees and other contributing family workers, in either the formal or the informal sector. A number of these enterprises did not keep accounts, thus limiting the possibilities of accurate measurement of income from self-employment.
80. It had been proposed that income from self-employment should measure the return to labour of the self-employed. This would mean deducting consumption of fixed capital as well as the return to investment from the enterprise gross mixed income. The participants agreed that, due to lack of records, the development of this model was primarily an academic exercise. A clear distinction between the return to labour and the return to capital was extremely difficult, if not impossible.
81. There appeared to be consensus on the fact that it would be more practicable if the measurement of income from self-employment focused on the gross mixed income of unincorporated enterprises.
82. Some participants questioned whether measured income from self-employment was a good indicator of welfare, considering the problem of income under-reporting (for tax avoidance and other reasons) in national surveys or inquiries. Some participants expressed their preference for a measure of expenditure as a proxy to income from self-employment, especially in the informal sector.
83. The question of the measurement of the imputed income of contributing family workers was a point of considerable discussion. A number of participants, including the Worker experts, were of the opinion that an imputed income should be estimated for this subgroup of self-employed persons, in order to assess their contribution to the activity of the enterprise. This would provide for a measure of self-employment income at the individual level. This measurement would be consistent with the measurement of income from paid employment.
84. Others expressed the view that it may be extremely difficult, on practical grounds, to apportion income among household members. Hence, it was decided that income from self-employment could be measured at the level of the enterprise defined as a single unit.
85. The participants agreed that the Bureau of Statistics should conduct additional research on different approaches in assessing income from self-employment, particularly in the difficult areas identified in the preceding paragraphs.
86. It was proposed by the secretariat that the measurement of income from self-employment be phased into national programmes of statistics over an extended period of time. Initial efforts should be made to identify two groups of self-employed persons.
87. The first group consisted of self-employed persons operating their enterprises with little or negligible capital input, who produced goods and services in a similar way to paid employees (such as those engaged in crafts or services, whether in the formal or in the informal sector). The income they received was mostly a return to their labour input and "gross mixed income" of the enterprise was a close estimate of net mixed income. When the self-employed worked in an individual capacity, the gross and net income of the enterprise represented the income of the individual; when they worked in partnership, each partner's income from self-employment was represented by a share of the enterprise mixed income.
88. The second group consisted of self-employed persons whose activities involved an identifiable amount of capital for the production and generation of income (such as professional workers in the formal sector or workshops in the informal sector). In this case, efforts should be made to quantify the amount of capital used for generating the income and to derive net mixed income, where relevant, by analytical measures.
89. The participants were in general agreement with this initial approach of quantifying income from self-employment in terms of gross or, where possible, net mixed income. This measure represented the return to labour, as well as entrepreneurship and investment of the self-employed.
90. There were extensive discussions on whether the income of the self-employed should be measured after deduction of that part of their mixed income retained and reinvested in the enterprise. However, the Meeting concluded that it was not appropriate to attempt to measure reinvested income, since this was part of income itself.
91. Self-employment activities could, over a given reference period, produce a financial loss instead of income. The Worker experts felt that, in principle, such losses should be reflected in the measurement of income from self-employment. It was agreed that losses should be evaluated as negative income.
92. The participants felt that the inclusion of employment-related social security benefits in the concept of income from self-employment was a minor issue. In a large number of countries, the self-employed were not covered by mandatory social security schemes and, when they were, the benefits may be low compared to those which were traditionally provided to persons in paid employment. However, the participants agreed that the treatment of employment-related social security benefits paid to the self-employed should be in line with the different views expressed on it in relation to income from paid employment.
93. The participants suggested that the methods to be used to record income from self-employment should follow, where feasible, either the procedures recommended by national tax authorities or the methods contained in the System of National Accounts. If this was not possible, the cash-flow method could be used.
Measurement issues
94. It was agreed that two basic statistical units were relevant to the measurement of income from paid and self-employment: the individual person and the enterprise. Individual persons should be covered regardless of age, gender and other socio-economic characteristics.
95. The Meeting identified a number of possible sources of data on income from paid and self-employment. Traditional surveys, such as labour force and other household surveys, appeared to be sources deserving consideration, in spite of the limitations inherent in this type of source (accuracy and reliability of income data, proxy response, memory recall, etc.). Where feasible, combinations of administrative records (such as tax and social security records) could prove useful. Other possible sources of data on income from employment included establishment surveys, agricultural surveys, informal sector surveys, surveys of small economic units and household enterprises. It was felt that there was a need to integrate various data sources to meet the requirements of the measurement of income from employment.
96. It was agreed that a long reference period, such as a full year, was appropriate to measure all the components of income from paid and self-employment. However, for data collection purposes, shorter periods, such as a month or a quarter, could be used, which would be aggregated to develop an annual estimate.
97. In addition to estimates of income on an annual basis, it was suggested that a measure of an hourly rate of income be developed. This measure could be based on estimates on income and on the corresponding volume of employment expressed in terms of hours of work and work duration.
98. Among the analytical measures of income from self-employment, it was suggested that, where relevant, countries estimate the imputed income of contributing family workers in unincorporated enterprises.
99. Regarding data collection, it was noted that details should be obtained on the various components of income from paid and self-employment.
Recommendations
100. The Meeting of Experts recommends to the Governing Body of the ILO that the topic of income from employment be included in the agenda of the Sixteenth International Conference of Labour Statisticians planned for 6-15 October 1998. The report that will be prepared by the secretariat for submission to the Conference should provide a clear definition of both concepts of income from paid and self-employment and should elaborate on their measurement. With regard to income from self-employment, the report should recognize that measurement may be particularly difficult for a number of categories of self-employed persons. The Bureau of Statistics should conduct specific research on different approaches in assessing this type of income, particularly in the difficult areas identified in this report, including the allocation of enterprise income among contributing family workers. With regard to income from paid employment, the Bureau of Statistics should undertake studies to examine the particular issue of the "imposed" compensation in kind in lieu of cash remuneration, in force in some transition countries, and develop appropriate guidelines. The draft resolution which will accompany the report to the 16th ICLS should be revised in line with the discussion of the Meeting as summarized in the present report.
Adoption of the report
101. The secretariat presented the draft report which contains the summary of the Meeting's discussions and draft recommendations. After examination of the report during two sessions, 17 and 23 October 1997, the experts adopted it.
102. The full report will now be presented to the Governing Body at its March 1998 session.