Saturday, 27 March 2010: 17:25
The measurement of racial discrimination crucially depends on the methodology used for identifying racial groups, a matter of no concern when the issue is gender discrimination, for example. A woman and a man can be easily distinguished through their personal characteristics wherever the analysis takes place. However when a survey is being conducted about prejudice, the researcher cannot objectively classify any individual, thus a subjective criteria must always be drawn.
In many countries racial segmentation is due to ethnical reasons. Nevertheless, even in this particular situation, there is no clear line to define who belongs to each racial group. Cultural and physical aspects are usually important to characterize identity, but often some of them are shared by more than one group and when it comes to the individual level any type of classification gets more complicated. In fact, in most democratic countries, a solution to this problem is simply to ask the survey´s participant to declare his/her racial group.
One of the main elements of discrimination is the existence of a segment of the population which benefits from the idea of superiority. Being so, racial discrimination can determine a trend in human behavior to declare oneself as belonging to the most favored group, which may define a self-selection problem. If this occurs, race self-selection must be taken into account to avoid biased results.
The objective of this study is to investigate the hypothesis of race self-selection using Brazilian data and then to determine to which extent it may bias results. To do that, a Heckman framework will be applied, discussing the strategies which must be used in light of race sample selection. This issue is of vital importance because, to our knowledge, this kind of sample selection has never been considered in the literature and depending on the way surveys are collected it may lead to severe biased outcomes.
Our empirical analysis uses the 1- Percent public use Microdata Sample of the 2000 Brazilian Census since it is the most recent available survey that links individuals to their respective households, an important information which will be used for obtaining instrumental variables.
In this way, the hypothesis of race self-selection was not rejected. After that, we estimate a Mincer equation, taking it into account. Results show that racial discrimination is underestimated in Brazil, reducing wages in 32.50 percent controlling for labor participation and race self -selection. If the latter is not considered, the effect of prejudice would decrease to 19.90 percent.
The end result of our analysis is that attention must be paid to the methodology of data collection because, depending on how it is obtained, the use of race variables may bias results. Besides that, it is clear that gender and race prejudice are an important issue to be faced by the Brazilian Society.
In many countries racial segmentation is due to ethnical reasons. Nevertheless, even in this particular situation, there is no clear line to define who belongs to each racial group. Cultural and physical aspects are usually important to characterize identity, but often some of them are shared by more than one group and when it comes to the individual level any type of classification gets more complicated. In fact, in most democratic countries, a solution to this problem is simply to ask the survey´s participant to declare his/her racial group.
One of the main elements of discrimination is the existence of a segment of the population which benefits from the idea of superiority. Being so, racial discrimination can determine a trend in human behavior to declare oneself as belonging to the most favored group, which may define a self-selection problem. If this occurs, race self-selection must be taken into account to avoid biased results.
The objective of this study is to investigate the hypothesis of race self-selection using Brazilian data and then to determine to which extent it may bias results. To do that, a Heckman framework will be applied, discussing the strategies which must be used in light of race sample selection. This issue is of vital importance because, to our knowledge, this kind of sample selection has never been considered in the literature and depending on the way surveys are collected it may lead to severe biased outcomes.
Our empirical analysis uses the 1- Percent public use Microdata Sample of the 2000 Brazilian Census since it is the most recent available survey that links individuals to their respective households, an important information which will be used for obtaining instrumental variables.
In this way, the hypothesis of race self-selection was not rejected. After that, we estimate a Mincer equation, taking it into account. Results show that racial discrimination is underestimated in Brazil, reducing wages in 32.50 percent controlling for labor participation and race self -selection. If the latter is not considered, the effect of prejudice would decrease to 19.90 percent.
The end result of our analysis is that attention must be paid to the methodology of data collection because, depending on how it is obtained, the use of race variables may bias results. Besides that, it is clear that gender and race prejudice are an important issue to be faced by the Brazilian Society.