Earnings Differentials : Understanding the Gender Wage GapTerm Paper Workshop
In a May 2014 document released by the Bureau of Labor Statistics it was reported that although almost as many women as men are in the workforce (52% and 70.2%, respectively), males earn a higher median salary than females regardless of job type or race. Additionally, the most current census data show that women’s average earnings in the United States are only 71.3% of men’s (US Bureau of Labor Statistics, 2014). This difference in earnings between men and women is not exclusive to the United States; however, it has been noted to be one of the largest of all the industrialized nations (US Bureau of Labor Statistics, 2014). Many researchers have attempted to explain why men earn more money than women. The gender wage gap has been partially attributed to women’s lack of upper management positions,differences in occupational choices (US Bureau of Labor Statistics, 2014), number of hours worked (US Bureau of Labor Statistics, 2014), education, age,position, job tenure, type of job, training, and organizational tenure.
Theories attempting to explain the gender wage gap have not been entirely successful. For example, it has been thought that women may want less money than men, but this has not been supported by the research (Szmania & Doverspike, 1990). Similarly, it was thought that women simply do not invest as much time into acquiring marketable capital such as education, training, knowledge, skills, and abilities, but empirical research does not support this (Szmania & Doverspike, 1990). The wealth of literature on the gender wage gap has provided much knowledge on the many variables that, when isolated, contribute to the earnings differential between men and women. The main purpose of this paper is to simultaneously examine these identified variables in order to get a better understanding of how they interact with one another in explaining the earnings differential.
Individual Level Variables
In a (2013) study, Ostroff and Atwater specifically examined the effect of age and gender on salary. They found that in their sample of over two thousand managers of various occupations that age and gender both significantly predicted salary. Furthermore, the interaction term between gender and age was also a significant predictor of salary after controlling for other variables. This indicates that the wage differential between males and females is larger at higher ages. Goldberg, Finkelstein, Perry, and Konrad (2014) replicated these findings in a sample of two hundred thirty-two MBA alumni from the same university. The women’s salaries did not increase much with age, although males’ salaries did. The gender by age interaction in these two studies suggests that simply controlling for age in investigating the gender wage gap may eliminate useful information on the different effect age has on males and females in terms of salary. As these studies suggest, there is a relationship between age and earnings, but this relationship may be different for males and females. Specifically, as men and women age, males’ earnings tend to increase at a faster rate than females’. Thus, being older tends to have a positive relationship with earnings for both males and females. However, the differences in salaries between men and women are greater at older ages versus younger ages, due to males’ age advantage in earnings. Like age, the effect of race has on earnings has historically been included in studies of salary research. It is commonly thought that in the United States, any non-white individual will earn less than their white counterpart.
It been suggested that salaries are lower for non-whites due to their minority status Research has suggested at one point that perceptions of others are heavily influenced by stereotypes based on characteristics such as race. Therefore, deflated earnings of non-white persons may be due to stereotypical beliefs of the person’s abilities rather than actual differences. Ostroff and Atwater (2013) also found that the manager’s race, measured as a dichotomous variable, was a significant predictor of salary in all three of their regression models. Miech et al. (2013) used a sample of over nine thousand individuals from the National Longitudinal Survey of Youth (NLSY) and found that being African-American significantly related to earnings as did gender, although their interaction was not significant. An overwhelming amount of the literature on earnings, nevertheless, has not found a relationship between race and earnings. For example, Mitra (2003) used a 1998 cross-sectional analysis of almost five hundred individuals in the NLSY database who hold a supervisory position. Race did not predict salary in the overall sample, nor in the analyses separated by gender. It is possible that these non-significant findings are due to the range restriction of the sample consisting of only those individuals who hold supervisory positions.
Toutkoushian (2010) did look at the different occupational levels, though, and found different relationships between race and salary. For the operating staff, race was a significant predictor of salary in one model only, and lost significance when the addition of the control for pay range was included. For all other analyses, including looking at a professional/administrative sample, overall and separated by gender, race did not significantly predict earnings. Ultimately, it seems that although race may affect earnings, gender has much more of an impact. Another individual-level variable shown to predict earnings is the average number of hours an individual works each week.
In Hinze’s (2000) review of the literature on the relationship between hours worked and wages, she suggests that, overall, working more hours increases earnings. However, Hinze also notes that there are gender differences concerning the relationship between earnings and number of hours worked. First is that even when hours are controlled for, men still out earn women. An example of this is if a man and woman both work an average of 40 hours a week, the man is likely to earn more money than the woman. However, when looking specifically within gender, there appears to be an advantage for women to work longer hours. For example, the difference in wages for two men who respectively work 40 and 45 hours a week is not as great as it is for two women with the same comparison. Thus, women may gain more than men by working longer hours as compared to their same sex counterparts. There are several studies in which earnings were found to be related to the number of hours worked. Toutkoushian (2010) regressed a variety of variables, including full-time status, on salary in a sample of 939 operating staff and 802 professional/administrative staff. He found that working full-time had a positive relationship with salary for the operating staff, but no significant relationship with salary for the professional/administrative staff. No gender differences were found between men and women. There is also evidence contrary to the conclusion that the number of hours worked positively affects earnings.
Other findings of this relationship indicate that number of hours worked may be higher for males, and thus leads to higher earnings. Specifically, Goldberg et al. (2004) used a sample of 232 MBAs and examined the effect of hours worked on salary. In a regression analysis, hours worked significantly predicted salary. No analyses were conducted based on gender; however, their non-directional correlational statistics reveal some interesting findings. Here, salary and hours were positively and significantly related. What does seem to matter, though, is that even if males and females work the same number of hours, men still out earn females. One example of this is found in the Levinson et al. (1994) study of a sample of 362 full-time school psychologists. They found that although there was no difference in the number of hours worked between males and females, the men in the sample still earned significantly more than the women. Other findings of this relationship indicate that number of hours worked may be higher for males, and thus leads to higher earnings. The implication with these findings is the suggestion that that the difference in salary between men and women may be partly due to women working fewer hours. For example, findings from a study on hours and earnings suggests that women may benefit more than men by working more hours. Hinze (2000) used a sample of 321 physicians who graduated from medical school between 1980 and 1990, and who were also married to another physician. Number of hours worked per week significantly predicted salary after controlling for several other variables. Further analyses indicate that this relationship is significant for women only.
Research has considered the effect of human capital on wage differentials and studied the difference of these effects between men and women for various occupations. Education level has been found to have some predictive capability for salary levels for women but not for men. At the administration level, the research signifies that a higher education is advantageous for women, but not for men. Yet, findings have not been entirely consistent. In Mitra's (2012) study that included many occupation levels, such as blue-collar, white-collar, professionals, managers, and highly skilled workers. Education, measured as the highest grade completed, was positively and significantly related to salary for males in each occupation, and for females in each occupation but blue-collar. Another study of blue-collar workers also found that education level is not advantageous for women. While at blue collar occupations, men's earning differential is commonly correlated with educational attainment differences, more robust studies also point to a differential earnings increase between men and women at higher-level occupations (Meier and Wilkins 2002)
Ostroff and Atwater (2013) examined the effect of education on salary and found similar results. Highest educational level achieved was measured with a 7- point scale ranging from 1 (some high school) to 7 (graduate degree). They found that in their sample of over two thousand managers of various occupations that education level predicted salary. Furthermore, the interaction term between gender and educational level was also a significant predictor of salary. This indicates that the wage differential between males and females is larger at higher levels of education. Overall, education increases earnings for males and females. The aforementioned studies do not indicate that males and females receive different amounts of education. However, in general it appears that males tend to earn more money with each educational attainment than women do, thus creating a wider gender gap as education increases.
Another human capital characteristic is the amount of years an individual has worked. Thus, research on the gender wage gap has also investigated the effect of number of years worked, either overall or full-time, to test its relevance on wages. One way this issue is conceptualized is by the amount of time individuals have been out of the workforce, known as career interruptions. The research on career interruptions has not been very prolific, and often the focus is on career interruptions due to maternity leaves The results of the research that has been conducted, however, are conflicting. Goldberg et al. (2004) found in their sample of MBA’s that there was no significant zero-order correlation between gender and number of times unemployed, and that being unemployed did not predict salary.
Domestic Predictors
Hinze (2000) validated this finding in a study on the effects of career interruptions due to child, measured in months, on salary. She used a sample of 321 physicians who graduated from medical school between 1980 and 1990, and who were also married to another physician. Mean differences in time off between males and females were significant, with females more likely to take time off for child rearing. In the regression models, time off was a significant predictor of salary in all models except for the last, in which hours worked per week was included. In this last model, number of children also lost significance when hours per week was added. Studies that have looked at the effect of having children on wages have shown that having children negatively affects the earnings of women only. Mitra (2012) examined the effects of marital status and number of children on earnings by occupational level. For the women, being married had a significant and negative effect on salary for the white-collar workers, professionals, and highly skilled workers. For males, being married had a significant and positive relationship with salary for the blue-collar, white-collar, professionals, and highly skilled workers. Having children or not was positively related to salary only for the males who were blue-collar workers. Conversely, having children was negatively related to salary for women managers. The implications of this study overall seem to suggest that marital status increases salary for males and decreases salary for women, regardless of occupational level. However, the presence of children only increases salary for males in low-skilled occupations, and only decreases earnings for women in managerial roles. In a follow-up study, Mitra (2013) examined these same variables with a sample of individuals holding supervisory positions. Marital status did not significantly predict salary, overall nor for either gender. Children did not significantly predict salary for females, yet it did for males, with more children related to higher earnings. These results suggest that marital status may not matter as much at the supervisory level, but presence of children benefits male earnings significantly more that female supervisory earnings. Dixon and Serron (2005) examined the effect of marital status and children separately by three organizational categories: private, government, and corporate.
Overall, there was a significant, positive relationship between being married and salary. Further analyses indicated that this significant relationship was only for the males working in the private sector, and females working for a corporate organization. There was also an overall significant, positive relationship between having children and earnings. However, differential analyses showed that this relationship is only true for the males, regardless of sector, and for females in private organizations only. In fact, for females in corporations, the relationship between children and earnings is significantly negative. This study indicates that the relationship between marriage, children and earnings is different for men and women and that the type of organization also moderates these relationships. Despite the different findings of these studies, there are some consistent themes. First is that whether the variable of interest is housework, marital status, or number children, there is evidence which suggests that gender moderates the relationship with earnings.
Second, since the number of hours of housework is not typically examined but has been shown to have a relationship with earnings for females, it is relevant to include this variable in earnings analyses as it may have an independent effect on earnings aside from the other domestic variables (Hersch & Stratton, 2007). Third, the majority of these studies show a trend that family life has a positive relationship with earnings for men and a negative relationship with earnings for females. Specifically, being married with children tends to increase earnings for males, while the opposite is true for females.
Organizational Characteristics
Occupational segregation refers to the tendency for men and women to work in different occupations. According to a review by Guy and Newman (2004) the difference in occupations is often regarded as the main reason why males make more money than females. The authors go on to question why predominantly female occupations pay less than predominantly male occupations. They suggest that the lowest paying occupations often require emotional work that is often thought of as a natural skill for women, and therefore goes unnoticed in terms of performance evaluations and compensation. Another similar theory, “status composition,” suggests that jobs that are held predominantly by females are viewed by the organization as having lower value. Often, female-typed jobs are devalued because they are considered to have lower skill requirements, which in turns lowers the pay and decreases opportunities for promotions.
A historical look at the segregation of occupations begs the question “Do women choose lower paying occupations or does a surplus of women in an occupation lower the earnings?” Research on the topic indicates that it is both, with an addition of societal factors working to facilitate the probability of females be attracted to and obtaining predominantly female jobs. For example, in an empirical investigation of perceptions of occupations, Hollenbeck et al. (1987) found that the women in the study were more interested in obtaining predominantly female occupations than the men, but the women also believed they had a higher likelihood of obtaining a job in a female occupation versus a male occupation. This study suggests that females may “choose” the lower paying jobs because that is all they think they can get. On the other hand, jobs may become lower paying when too many females choose them. For instance, when jobs are predominantly held by women, they are perceived as being less prestigious and deserving of lower salaries than those predominantly held by men.
Research also suggests that organizational practices play a role in occupational segregation. Differences in hiring, promotions, and job placement lead women to be allocated to lower-paying jobs than men . Miech et al. (2013) longitudinally examined the obtainment of human capital, or differential investments, between men and women in an attempt to explain occupational segregation. They used two data sets, which spanned 11 years and 16 years respectively. The major findings from both of these studies is that at each data point, the males had higher levels of occupation income, yet the females had higher levels of occupational education, defined as one or more years of college. The authors suggest that the amount of education may not be as important as the type of education. However, they also acknowledge that as soon as women start getting the “right” education, i.e. ones that lead them to predominantly male occupations, the wages may become depressed by the saturation of females in the field. Whether occupational segregation is due to individual choice in occupation selection, societal influences and pressures, and/or biasing organizational practices, occupational differences between males and women may account for a large amount of variance in the gender wage gap. Nelson and Bridges (1999) found that the percentage of females in an occupation and one’s gender better predicted an individual’s earnings than average market pay rates. Gibelman (2013) also looked at the relationship between percentage of females in an occupation and earnings. She found that as the number of women in an occupation decreased, the average salaries for the occupation increased. Finally, using 1987 data from the from the Panel Study of Income Dynamics (PSID), Maume (1999) found a gender interaction between sex and the gender composition of occupations such that the percentage of males in the occupations significantly increased hourly earnings for males only.
Conclusion
In this paper I have described the numerous variables that have been shown by previous research to predict earnings. For each variable, I described the extent to which gender moderates the relationship between the variable of interest and earnings. As outlined, some variables have a positive relationship with earnings for men and women such as age, hours worked, and education. Other variables may predict earnings for one gender only. And finally, gender moderates the relationships between earnings and other variables. One example is that being married tends to have a positive relationship with earnings for males and a negative relationship with earnings for females. Gender also moderates the relationships to earnings for variables such as number of children and organizational sector.
References
Clark, L.H. (1993). Speaking of business: The pay gap narrows-slowly. Wall Street Journal A, 6:4.
Dixon, J., & Serron, C. (2005) Stratification in the legal profession: Sex, sector, and salary. Law& Society Review, 29, 381-412.
Goldberg, C. B., Finklestein, L. M., Perry, E. L. and Konrad, A. M. (2014). Job and industry fit: The effects of age and gender matches on career progress outcomes. Journal of Organizational Behavior, 25, 807-829.
Hersch, J. & Stratton, L. S. (2007). Housework and wages. Journal of Human Resources, 37, 217-229.
Hollenbeck, J. R., Ilgen, D. R., Ostroff, C., Vancouver, J. B. (1987). Sex differences in occupational choice, pay, and worth: A supply-side approach to understanding the male-female wage gap. Personnel Psychology, 40, 715-743
Meier, K. J., & Wilkins, V. M. (2002). Gender differences in agency head salaries: The case of public education. Public Administration Review, 62, 405-411.
Miech, R. A., Eaton, W. & Liang, K. (2013). Stratification over the life course: A comparison of occupational trajectories across race and gender during the 1980s and 1990s. Work and Occupations: An International Sociological Journal, 30, 440-473.
Mitra, A. (2003). Access to supervisory jobs and the gender wage gap among professionals. Journal of Economic Issues, 37, 1023-1045.
Mitra, A. (2012). Mathematics skill and male-female wages. The Journal of Socio-Economics, 31, 443-456.
Maume, D. J. (1999) Occupational segregation and career mobility of white men and women. Social Forces, 77, 1433-1459.
Ostroff, C., & Atwater, L. E. (2013). Does whom you work with matter? Effects of referent group gender and age composition on managers' compensation. Journal of Applied Psychology, 88, 725-740.
Szmania, J. M., & Doverspike, D. (1990). A test of the differential investment hypothesis applied to the exploration of gender differences in reactions to pay. Journal of Vocational Behavior, 37, 239 – 250.
United States Bureau of Labor Statistics (2014, May). Women in the labor force: a databook. Report No. 1049. Washington, DC: Government Printing Office. Retrieved from
http://www.bls.gov/opub/reports/cps/womenlaborforce_2014.pdf
