Unemployment Rates in Brazil
This article should not be considered as an analysis of the causes and effects of employment in Brazil. Rather, we want to discuss certain aspects of the definition of employment and unemployment, and we happened to use Brazil to illustrate.
Our reference database is the 2003 TGI Brazil study. This is a survey of 10,623 persons between the ages of 12 to 64 years old who were interviewed during 2003. This is not a full national study, as it is limited to the major urban areas in Brasilia, Belo Horizonte, Curitiba, Fortaleza, Porto Alegre, Recife, Rio de Janeiro, Salvador, Sao Paulo, the interior of Sao Paulo state and the interior of the Sul/Sudeste region. The projected population in this sample universe is 57,034,000 persons 12 to 64 years old.
These survey respondents were asked about their employment status. A projected total of 32,302,000 persons responded that they were presently employed. Now, this does not mean that the employment rate is 100 x 32,302,000 / 57,034,000 = 56.6%. That is because that the total survey universe will include people who are not old enough (e.g. a 12-year-old girl who would be breaking child labor laws if she worked), people who are attending school (e.g. a 20-year-old university student), people who are retired (e.g. a 64-year-old former firefighter), people who are disabled (e.g. a blind person), people who have no intention to work (e.g. a multimillionaire heiress) and so on. The employment rate should be restricted to set of people who are in the labor force.
By convention, the active labor force is defined to be those who are currently employed plus those who are not currently employed but are actively looking for work, of which there are 4,023,000 persons in this case. Thus, the employment rate would be 100 x 32,302,000 / (32,302,000 + 4,023,000) = 100 x 32,302,000 / 36,325,000 = 88.9%. Conversely, the unemployment rate is 100% - 88.9% = 11.1%. That is the conventional definition.
Periodically, news reports carry some apparently paradoxical stories about how the unemployment rate is going down even though the number of job layoffs is going up and the number of newly created jobs is also going down. Under the framework so far, this would seem to defy quantitative logic.
It so happens that the convention includes a separately category known as the 'discouraged worker.' These people desire to find work, they have looked for a long time without success and they have given up the search process. Under the convention definition, these people 'might' have gotten a job if they tried, and so they are not counted.
In our present case study, there were 682,000 persons who are currently unemployed but are not actively looking for work. If these people were included in the base, the unemployment rate would have been 100 x (4,034,000 + 682,000) / (4,034,000 + 682,000 + 32,302,000) = 12.7%, of which 1.8% are 'discouraged workers.'
It may happen that the economic prospects may be so bleak that many people become discouraged, and the net result may look as if the unemployment is coming down. Conversely, the economy may be improving but more 'discouraged workers' are re-entering the job search and therefore it may look as if unemployment is going up. The conventional formula to calculate the unemployment rate can therefore lead to such seemingly paradoxical results.
In the next slide, we show a comparison of the profiles of the two unemployed categories. Since the incidences are different, we have re-expressed everything as an index with respect to the total incidence. For example, the total incidence for 'discouraged workers' is 1.8%; since the incidence among males 12-19 is 5.2%, the index is 100 x 1.8 / 5.2 = 281. For socio-economic level, the profiles are similar. Unemployment affects the lower classes more than the upper classes. By age/sex groups, the profiles are sharply different. Among males, the relative incidences of unemployed inactives are much higher among the youngest (who have difficulty finding their first job) and the oldest (who have difficulty finding new jobs after being laid off); among the females, the child-bearing age between 25-34 years shows the highest relative inactivity. For both males and females, the relative incidences of unemployed actives are highest among the youngest. To the extent that the conventional calculation excludes a certain segment of the theoretically eligible population, we are reminded by these data that the exclusion affects certain socio-economic and demographic groups more than others.
(data source: 2003 TGI Brasil)
Next, we look at the definition of the currently employed workers. The convention is such that anyone who works is considered employed. If we break down the 36,225,000 employed persons, 79.5% of them worked full-time (defined as 30 hours or more per week) and 20.5% of them worked part-time (defined as less than 30 hours per week). In the next chart, we show the profiles of full-time and part-time workers, with all numbers being expressed as indices as before. Indeed, rich people have less need to put in more than 30 hours per week at work. Among males, the highest incidence of part-time employment is for teenagers, who may be working after school; among females, the highest incidences of part-time employment is among the youngest and oldest. Overall, women are more likely to work part-time than men. We bear in kind that the choice of full-time and part-time is sometimes a matter of individual choice (e.g. students working after school, or young mothers who need to take care of children), but it may also be by force of circumstance.
(data source: 2003 TGI Brasil)
To illustrate that not all employment are equal, we show the indexes of full-time and part-time employed according to monthly personal income. The majority of part-time workers will make less than R$350 per month, whereas more than two-thirds of all full-time workers will make more than that. Thus, it is not sufficient to consider the employment rate alone, since there may also be an underlying shift between full-time and part-time work.
(data source: 2003 TGI Brasil)
This little exercise about some underlying issues on the definition of employment/unemployment rates. These are by no means the only issues involved, but it certainly illustrates the need to be wary of relying solely on a single statistic as if it accounts for everything that is going on.
(posted by Roland Soong, 12/06/2003)
(Return to Zona Latina's Home Page)