The Gender Gap in Earnings Among Teachers: The Case of Iowa in 1915

ABSTRACT

This paper draws on the 1915 Iowa State Census Report to decompose the gender gap in earnings into explained and unexplained parts. A novel feature is that the decomposition is performed not only at the mean but also over the entire distribution of earnings. In addition, an entire state, rather than a few cities, is considered. This paper finds that at least 25.6 percent, and probably more, of the gap is unexplained by the main observable characteristics at the mean. More interestingly, the unexplained part grows moving up the distribution of earnings, which indicates the possibility of a glass-ceiling effect for women. Results provide new insight into gender wage gaps among the highly educated, theories and empirical analysis in labor economics, and quantification in the history of education.

Keywords:

• Earnings discrimination
• economic methodology
• economic history
• feminist economics
• occupational segregation
• labor supply

JEL Codes:

• I22
• J16
• N32

NOTES ON CONTRIBUTOR

Kitae Sohn is Assistant Professor at Kookmin University in South Korea. He holds a PhD in Economics from State University of New York at Albany and a MSc in Economic History from the London School of Economics. In 2012, his articles were published in Bulletin of Economic Research, Education Economics, and Journal of Social History. In 2013, more articles were published or in press in Historical Social Research, Developing Economies, Singapore Economic Review, Labor History, Journal of the Asia Pacific Economy, Journal of Biosocial Science, and Economics Letters.

ACKNOWLEDGMENTS

Early drafts of this paper have been presented at the 2012 Annual Meeting of the Korean Association of Applied Economics and the 2012 Korean Economic Association International Conference. I am grateful to Joo-Young Kim and participants at the meetings for helpful comments. I also thank an Associate Editor and five anonymous reviewers for constructive suggestions and comments.

Notes

1. When administrators and teachers are considered together to estimate the earnings gap, precise estimation requires that one take into account more factors such as composition effects, promotion probabilities, and different duties.

2. The three cities refer to Davenport, Des Moines, and Dubuque. The urban sample accounts for 5.5 percent of the population in the three cities, and the rural sample accounts for 1.8 percent of the population in counties without large cities.

3. Observations with missing values were also included in the analysis by creating a dummy for the observations and setting the missing values to zero; the results (not shown) change little.

4. Among 300 women whose marital status was identified, 288 women were single, eleven women were married, and one was a widow. Informally, the marriage bar began to be lowered when the Second World War created manpower shortages; formally, lawsuits against this form of discrimination (such as Houghton vs. School Committee of Somerville [MA] in 1953) contributed to the elimination of the rule.

5. Hourly wages are better suited for estimating wage gaps, but they are unavailable in the data.

6. As one reviewer suggested, I also experimented with normalized weights, which were calculated by dividing the survey weight of each unit by the average survey weight of the sample. The results are identical with those estimated with sampling weights. In addition, estimations without any weights produce similar results.

7. The possibility of a type II error for the insignificance cannot be dismissed.