Since the 60s, the IBGE have recognized the need to conduct a regular household survey to be able to better understand the changes in the country’s social, de- mographic and economic development. IBGE’s main national household survey PNAD1, which started in 1967 as a quarterly survey, became annual in the late
70s. However, PNAD did not provide information on short term changes in the labour market. Hence, IBGE developed its labour force survey, first implementing it in 1980 (IBGE, 2001a). The main objective of the PME is to investigate the characteristics of the Brazilian labour force, producing monthly indicators that aid the evaluation and planning of the country’s socio-economic development (IBGE, 2009). The PME is a labour force survey comparable to the UK Labour Force Survey (UK-LFS) and the U.S. Current Population Survey (CPS).
Brazil is the largest country in South-America, and according to the IBGE the estimated population in the year 2005 was about 186 million inhabitants. Ge- ographically, the country is divided into 5,560 municipalities. These municipalities are grouped into 26 Federation Units (or States), and one Federal District. These states are grouped into five great regions: North, North-East, Centre-West, South- East and South. Some municipalities are also grouped to form the 28 Brazilian metropolitan regions.
Recognizing that the core of the Brazilian labour market is located in the metropolitan areas, the PME was initially designed to cover the nine metropolitan areas that existed at the time the survey was introduced, and the Federal District. However, due to budget constraints, the survey was first implemented in two of these regions and two years later it was covering six metropolitan regions: Re- cife and Salvador both located in the North-East Region; Belo Horizonte, Rio de Janeiro and S˜ao Paulo located in the South-East and Porto Alegre located in the South of Brazil. These are still the six metropolitan regions currently covered by the survey.
PME has been one of the main sources of labour market indicators in Brazil. Since it was first implemented, the PME survey methodology required some re- visions. The first revision was in 1982. It was concerned with changes in the conceptualization of work and aimed to relate to the questionnaire of the national household survey PNAD. Later, in 1988, the survey had its second revision, where
only the sample size was reduced (Silva and Moura, 1988). This second revision was justified as it improved the quality of the field work. Before this revision, a listing process was carried out every two years with a sample of new households found in selected enumeration areas being added to the original sample. However, the number of interviewers remained the same or even reduced over the years. This resulted in an increasing number of incomplete interviews. Silva and Moura (1988) showed that this survey revision reduced the rate of non-response from around 24% to 5%. Later, in 1993, a similar revision was performed again in order to reduce the non-response rate.
In 1996, besides the PME, there were other surveys investigating the labour force being conducted by other institutes of research in Brazil. In an effort to improve the utility of the indicators provided by the different surveys, the Min- istry of Labour proposed the unification of such surveys. For this purpose IBGE started the latest revision of PME. No consensus was reached on the unification of such surveys which differed mostly on the definition of the main labour market indicators and design. However, IBGE continued with the revision process recog- nizing that the PME survey should be updated to be able to capture the latest changes in the Brazilian economy (IBGE, 2002). The revised survey was only fully implemented in 2002. The main modifications concerned the definition of certain variables following recommendations from the International Labour Organization (ILO) to be applied to general labour force surveys. In the PME this involved modifications of the sampling design, the questionnaire ordering and the regional coverage. The revised PME was reduced to cover only the urban areas of the six already included metropolitan regions. Due to important methodological changes in this latter revision, the historic series initiated with the first implementation of PME, was discontinued. However, IBGE still makes available both series.
IBGE is the first user of the PME data: it uses the PME mainly to pro- duce monthly reports comparing the monthly employment indicators and the gross flows in and out of employment. It also produces technical reports such as IBGE (2001a,b, 2002, 2003, 2009). Some recent studies conducted by IBGE using the PME data have been: on the increasing participation of women in the labour mar- ket and their allocation as the head of the household (IBGE, 2006b); on the profile of domestic workers and their increasing participation in the labour force (IBGE, 2006c) and also on the increasing participation of the population aged over 50 in the labour market (IBGE, 2006a).
Other users besides the IBGE are those that implement more elaborate anal- ysis of the PME data. However, the majority of the studies conduct cross-sectional
or pooled cross-sectional time series analyses, while very few conduct longitudinal analyses (Corseuil and Santos, 2002). One example of the use of the PME explor- ing its longitudinal component can be found in Sedlacek et al. (1989). This study used data from the PME before the latest revision to investigate labour market mobility in a short period of time. The problem of matching individual records was raised in this study. Ferrao (2002) is also another example of the use of the PME data before its latest revision. This study used the multilevel modelling ap- proach to investigate the participation in the labour market. Other studies, also utilizing the PME data before the latest revision, are: Lemos (2002, 2006), both used old PME with matching difficulties mentioned in the former; Schwartzman (1999) that mentioned the reformulation and implementation of the question for skin colour; and Barros et al. (2000) which used pooled time series analysis. The lack of studies exploring the longitudinal component of the PME might be justified by the difficulty in matching individual records to form a proper longitudinal data set. This issue is discussed later on in this chapter.