The present study examines the impact of physical attractiveness and intelligence on the determination of wages for both current and potential employees. Physical attractiveness has already been shown to serve as a signal for intelligence – the so-called ‘beauty premium’ (Hamermesh, 1994). This research hypothesizes, however, that once the intelligence of a (potential) employee is established, the effect of the ‘beauty premium’ is minimized or may disappear entirely. Wages are thus determined based more on intelligence than on rewards for natural physical endowment. Physical attractiveness, then, may be an important but not a deciding factor in the determination of wages, or in climbing up the career ladder.
Ever since Mincer?s (1974) pioneering work on ‘Schooling, Experience and Earnings’ many economists have tried to identify the most relevant factors determining wages. Identification strategies have included regressing wages on various sets of determinants including gender (Blau et al., 1996), health status (Schultz, 1985), race (Heckman, 1998), and intelligence (Plug, 2000). Some scholars have even analyzed the effect of height (Persico et al., 2004) and obesity (Sargent et al., 1994) on the level of wages. But of more importance for the proposed project is the fact that more than 30 years ago psychologists confirmed that one?s physical attractiveness is advantageous to securing employment. Inspired by this result, economists began to analyze the effect of physical attractiveness on the labor market more generally. The aptly-called ‘beauty premium’ (Hamermesh, 1994) was found to function as a signal that indicates intelligence; the more attractive an employee is deemed to be, the more intelligent she is assumed to be. The ‘beauty premium’ thus increases the probability to get hired, on the one hand, and increases the offered wage once it comes to employment, on the other. This finding, that physical attractiveness has a positive effect on an individual’s labor market position, has been shown to exist across all types of industries. It has also been the subject of more extensive investigation since then (e.g. Averett and Korenman, 1996; Biddle and Hamermesh, 1998; Mulford et al., 1998; Bowles et al., 2001), which has repeatedly confirmed that beauty indeed does have a positive impact on wages.
Although the consensus seems to be that the ‘beauty premium’ is an undisputed factor in determining labor market position and individual wages, it is still unclear what roles other factors play in determining wages, and what impact these might have on the importance of the ‘beauty premium’. The question which forms the basis for the proposed research, therefore, is whether the ‘beauty premium’ is robust enough to stand up to other determinants of wages, namely intelligence. That is, given that the ‘beauty premium’ exists, can its impact be diminished by introduction of another variable, such as intelligence? Can intelligence outweigh the ‘beauty premium’, and thus have a greater impact on the determination of wages? If intelligence is found to have this greater effect, this matters for investments in human capital: Schooling enhances natural intelligence, and if it can be found that intelligence takes precedence over the ‘beauty premium’ in the labor market, then it is worth investing in schooling to promote an ‘intelligence premium’. If, however, physical attractiveness is found to have a greater impact on the determination of wages, this might matter for investment in beauty. Thus, it might be more worth investing in physical attractiveness than in schooling. This, however, would contradict the existing literature (e.g. Mincer, 1974; Averett and Korenman, 1996; Hamermesh, 1998).
To conclude, there is no study that estimates the effect of intelligence and beauty on wages simultaneously. This is a very important gap in the existing research which my proposed research seeks to fill.
The main objectives of the proposed research project are as follows:
To estimate the effect of physical attractiveness on wages
To estimate the effect of intelligence on wages
To design a new model of earnings
To analyze whether the impact of intelligence outweighs the ‘beauty premium’
Different strategies are available for regressing wages on different sets of determinants. In order to estimate the effect of intelligence and physical attractiveness on wages properly, I plan to use OLS estimation, as it should provide the most economically significant results, and has the advantage that results are very straightforward to interpret, which facilitates the analysis.
The first step is to apply the standard form of Mincer?s (1974) wage regression to determine whether my results would be in line with those in the literature,
log (waget) = ? + ?1 (educt) + ?2 (expt) + ?3 (expt)? + ?t (1)
where educ denotes years of education and exp the accumulated labor market work experience. ? is the level of non-qualified wage, ?1 the rate of return to education, and ?t a mean zero residual with E(?t) = 0. I refer to ?1 as the return to education and it identifies the marginal effect of education in percentage on log wages. Likewise, ?2 and ?3 are those effects that correspond to the return to experience.
The second step adds the beauty factor (physical attractiveness) to the model, to make out whether the ‘beauty premium’ indeed exists and to what extent it affects wages:
log (waget) = ? + ?1 (educt) + ?2 (expt) + ?3 (expt)? + ?4 (beautyt) + ?t (2)
Here, (beauty) denotes the measure for the physical attractiveness of the individuals.
The third step entails analyzing the impact of intelligence on the determination of wages. For this, I will introduce a variable for general intelligence. A general factor for intelligence that is industry independent is needed, so as to avoid problems with industry-specific knowledge.
log (waget) = ? + ?1 (educt) + ?2 (expt) + ?3 (expt)? + ?4 (intelligencet) + ?t (3)
Finally, I will introduce into my model, an extension of Mincer’s (1974) model. In my model wages are not only determined by years of education and work experience but also by physical attractiveness and intelligence. As a result, all the mentioned determinants are merged in one single regression with the purpose of testing the hypothesis, namely whether the ‘intelligence premium’ outweighs the ‘beauty premium’:
log (waget) = ? + ?1 (educt) + ?2 (expt) + ?3 (expt)? + ?4 (beautyt) + ?5 (intelligencet) + ?t (4)
If the model fits my cross-sectional dataset of heterogeneous individuals, I can then confirm my hypothesis and tackle the question: What is the marginal effect of beauty after accounting for another cause of variation in earnings, namely intelligence? Of course, to avoid multicollinearity, I will also control for other potential factors, such as gender, race and health status.
Moreover, to determine whether the ‘beauty premium’ is of higher significance for one particular gender, I split the sample and run two separate regressions, one for the female sample and one for the male sample. It is expected that the ‘beauty premium’ is of more importance for women than for men, however, once physical attractiveness and intelligence are included into the regression the effect of the ‘beauty premium’ will be reduced.
The data come from the National University of La Plata (Universidad Nacional de La Plata), Argentina and was collected by Dr. Martin Teraz, Professor for Labor Economics. The data was drawn from a random sample of 929 individuals in the area of –‘Gran La Plata’–, including the cities of La Plata, Berisso, and Ensenada. –‘Gran La Plata’– is inhabited by 600 000 people in the vicinity of Buenos Aires. The broad household survey was composed of 70 questions, including an IQ test as a measure of general intelligence. The survey provides data on the respondents? physical attractiveness as well as on the usual labor-market variables of interest for economists. The survey has the advantage, for the purposes of this research, of including substantial background information on the respondents. In order to get a significant and unbiased parameter for beauty, 10 randomly chosen individuals rated independently the respondents? physical attractiveness on a 10-point scale, which range from homely (1 point) to beautiful (10 points). The table below summarizes the main variables with their basic characteristics.
Main Variables
Variable
Obs.
Mean
Std. Dev.
Min.
Max.
gender (1 stands for males)
929
0,4477
0,4975
0
1
Age
929
37,9300
15,9600
18
81
years of education
929
13,3100
4,0800
0
18
monthly wage
545
1019,5400
780,9800
100
7700
general factor of intelligence
929
-1,59E-09
0,7825
-1,3620
1,6310
physical appearance
929
6,0139
2,2535
1
10
Although the data includes information on gender, age, years of education, monthly wage, a general factor for intelligence, and a measure for physical attractiveness, it does not contain a variable for ‘years of experience’, and hence age is used as a proxy for ‘years of experience’.
The dataset does not provide a complete distribution of wages conditional on the physical attractiveness of the respondents; this might be due to the fact that some of the respondents do not work. Of course, those individuals do have a shadow wage, i.e. a reservation wage for which they would be willing to work, but this wage is not known. This lack of data may cause a selection bias which could result in a bias of my estimates, if the decision to work is somehow correlated with physical attractiveness. In order to control for this possibility I will introduce the Heckman selection model (1979).
This research will extend existing studies on the determination of wages. It is expected that beauty will be shown to indeed be a significant factor determining wages, in line with previous research. It is also expected that both determinants, physical attractiveness and intelligence, will affect wages positively. Where this research differs, however, is in the expectation that once these two factors are included into the regression, the ‘intelligence premium’ will outweigh the ‘beauty premium’ and be shown to ultimately have a stronger effect on the determination of wages.
