Pay statistics are one of the most difficult topics for data collection. They can be derived from three major types of sources: establishment surveys (including administrative reports submitted by establishments/enterprises to the national authorities and assimilated to “surveys”); labour force/household surveys; and administrative sources (mostly social security records). Each type of source has its own characteristics and provides certain types of data.
In household-based surveys, which are the only source that can guarantee total worker coverage in most countries, including coverage of self employed workers, it is a sensitive subject to ask. The quality of the responses is expected to be lower than in establishment based surveys, where information is obtained from payrolls. In a growing number of countries, however, wages estimates based on household-based surveys for employees in the formal sector are consistent with establishment-based estimates. The primary objective of labour force/household sample surveys (LF/HHS) is to provide information on the levels and trends of employment, unemployment and sometimes underemployment. They sometimes give information on earnings and hours of work from the main and secondary activities of individual members. Pay statistics can be disaggregated by sex as well as by other worker characteristics.
Regardless of the limitations of household-based surveys, given their ability to cover the entire working population, to provide individual worker information on a multitude of subjects and their overall measurement flexibility, they are used to measure not only wages (or income from employment) but also total household income. Household based surveys can provide wage statistics by sex, to calculate the gender pay gap; by occupations, to measure wage inequalities by occupation; by formal/informal sectors and by other variables, including union membership; these surveys can produce statistics on the distribution of wages (the number of workers by different wage levels) to obtain wage inequality indicators.
Establishment-based surveys obtain information from payroll data and refer to gross cash remuneration. They are considered to be the source that will produce the most consistent and error-free information on wages. However, they tend to cover regular employees in formal sector establishments, leaving out the majority of workers in developing countries. They do not include casual workers, family members or the informal sector. Establishment based surveys will not produce representative statistics for the whole working population, but only for a subset of it. In principle it is possible to derive data by sex. However, in a number of countries, these surveys provide data for both sexes combined only, either because the accountancy/financial rules in force in establishments do not require to keep more detailed records, or because it is considered too much of a burden for employers to provide details by sex – with, as a consequence, the risk of high non-response rates to the survey. In addition, most establishments provide information on regular earnings that are received each pay period in cash. They exclude irregular payments that are not received each pay period but may be received regularly (such as the 13th month salary) and payments in kind, which can be important elements of earnings as defined in international standards.
Ideally, both types of sources should be used to measure wages (income from employment), but in developing countries, priority should be given to including questions on wages in household-based surveys.
The ILO has been regularly compiling two types of statistics of average wages (wage rates and/or earnings): (i) by branch of economic activity and in manufacturing by industry group, and (ii) by occupation, for a pre-defined list of 159 occupations within 49 industry groups. These statistics are published, respectively, in the ILO Yearbook of Labour Statistics and in Statistics on occupational wages and hours of work and on food prices – October Inquiry results, and available in LABORSTA on line since 1969.
Whenever possible, these statistics are collected by sex. However the possibility of obtaining wages statistics by sex is largely influenced by the type of source of the data.