Printer Friendly

Entry level salaries of academic economists: does gender or age matter?


Since the mid 1970s earnings differentials in academic labor markets have been extensively investigated with particular attention focused on salary differentials between males and females. The leading studies use similar methodologies but differ in terms of the time period analyzed, data sources and whether a single academic discipline or multiple disciplines are considered. For example, Johnson and Stafford |1974~ investigate several academic professions in 1970. Jusenius and Scheffler |1981~ consider a single discipline (economics) in 1973 and include a cross section of universities and professorial ranks as well as race. Several investigators including Katz |1973~, Hirsch and Leppel |1982~, Rickman |1984~ and Raymond, Sesnowitz and Williams |1988~ focus on a single college or university and include diverse academic disciplines. The latter two studies raise the question of whether gender continued to matter in academic labor markets in the 1980s and reach opposite conclusions.

A consistent finding transcending early studies of academic labor markets is that earnings differentials are smallest between men and women who are at the beginning of their careers. For example, Johnson and Stafford, in a widely cited study, state: "The most important finding is that the salaries of females start out at not much less than those of males (4 to 11 percent in the six disciplines in our sample) and then decline..." |1974, 901~. This paper investigates entry level salaries in a single academic discipline by considering a cross section of newly hired economists. The primary source of labor market information is a unique survey of entry level salaries in the 1987-88 academic year. It might seem that by the late 1980s gender differentials should have disappeared at the entry level, but like Johnson and Stafford's |1974~ earlier result, our survey reveals newly hired females received $1060 (3.4 percent) less than males. Our principal purpose is to determine whether this gender differential and the effects of age are significant once other determinants of earnings are taken into account. For example, we consider and control for terminal degree status, the quality of the Ph.D. granting department, field of specialization, highest degree offered by the hiring department, differences in cost of living across areas, and other institutional factors that can influence beginning entry level salaries.

We first describe the data set developed primarily from a survey, which is used to estimate entry level earnings equations. We also briefly compare our survey to other surveys providing information on market outcomes for economists. We then present estimates of the determinants of entry level salaries and discuss the influence of gender, age and other determinants of earnings.


A total of 469 U.S. economics departments were surveyed in the summer and fall of 1987 with 258 departments responding (55 percent).(1) The 1987-88 academic year salary of each new hire was requested of each department. Demographic information (age, sex) and human capital characteristics (Ph.D. granting university, primary field of specialization and degree status) were also requested for each new hire. In addition, the survey included questions on whether the hiring college/university is privately or publicly supported, the administrative reporting channel of the economics department, highest degree offered and number of entry level persons hired. The identities of the hiring department and the university at which the newly hired economist did his or her graduate work are used to construct two measures of quality. The quality of the department where the Ph.D. work was done is the first measure and the quality of the hiring department is the second.(2)

The survey was designed to provide microdata on individual entry level economists for the purpose of estimating earnings equations. In an effort to maximize responses the questionnaire was intentionally brief; only questions directly relevant to estimating the earnings equation of entry level economists were included. Our survey differs in several important respects from two other surveys of market outcomes, which have other purposes and provide quite different information. We considered using data from the American Economic Association's (AEA) Universal Academic Questionnaire, which is mailed annually to approximately one thousand colleges and universities. However, the AEA's questionnaire provides only departmental average data by rank, not microdata, and the response rate is typically quite low.(3) Further, the questionnaire does not ask questions about newly hired economists, focusing instead on all male and all female economists at different faculty ranks. In contrast, the so-called "Stromsdorfer Survey,"(4) sponsored by the AEA Chairperson's Group(5) does collect microdata on new hires. The Stromsdorfer Survey is sent annually to approximately one-hundred of the largest and most prestigious departments and collects extensive information, which is shared with members of the Group and others who request it. The purpose is to provide market information to departments that play important roles on both the demand and supply side of the entry level market. While containing extremely useful information, the Stromsdorfer Survey does not collect the detailed microdata required to estimate earnings equations. For example, the salaries of individual economists are not a part of the survey nor is the sex or age of the new hire.(6) For the 1987-88 academic year the seventy-nine departments responding to the Stromsdorfer Survey (in the fall of 1988) reported 138 entry level hires and a mean salary of $34,669, with a standard deviation of $2209. This compares to our sample of 258 responding departments, which hired 268 entry level economists, paying a mean salary of $31,523 with a standard deviation of $3874.

The descriptive statistics of our survey are summarized in the appendix.(7) Some of the more salient are worth emphasizing. The reported salaries ranged from a high of $42,000 to a low of $20,000. Of the total 268 new hires, 211 were male (78.7%) and 57 were female (21.3%). The average salary for males was $31,748 while that of females was $30,688, a $1,060 difference. The mean salary of those accepting positions in departments with a quality ranking based on research output was $33,198 while those taking jobs in unranked departments averaged $28,370. Economists accepting positions in business schools were paid on average $744 more than those in economics departments housed in other administrative units. While of interest, comparisons of sample means of the sort reported above cannot, of course, answer questions about the determinants of market outcomes. The next section specifies and estimates a standard earnings equation model to assess the effects of gender and age on entry level salaries, while holding other influences constant.


The extant literature suggests that the appropriate functional form for estimating an earnings function for entry level economists is of the following semilog form:

|Mathematical Expression Omitted~,

where |E.sub.i~ denotes the salary of the ith entry level economist, |X.sub.j~ refers to earnings determinants, |b.sub.j~ represents the parameters to be estimated, and |U.sub.i~ is an error term with a zero mean and constant variance.

The earnings determinants include the variables shown in Table I, several of which require comment. Since living costs vary widely across the dispersed geographical areas of responding departments we use INCOME as a proxy for local cost of living indices, which are unavailable. INCOME is measured by per capita income in the county in which the hiring department of the college/university is located. Of course, INCOME is not a perfect proxy for costs of living differences across areas. For example, INCOME also reflects area specific quality of life considerations. Thus, as a yardstick of the cost of living INCOME contains some measurement error, but in the absence of more TABULAR DATA OMITTED reliable indices we use it as a proxy.(8) Since ln |E.sub.i~ is the dependent variable in equation (1), we use the logarithm of INCOME as our proxy for the cost of living. The number of NEW HIRES is included because it may reflect the intensity of institutionally specific needs and the financial commitment to hire new economists.

Hirsch et al.'s |1984~ measure of research productivity per faculty member in 243 departments is used as an index of quality.(9) Of course, there are other dimensions to departmental quality, but research can be quantified and is widely thought to be an important factor influencing the entry level market for economists. We identify and use two quality variables, QUALITY-PD and QUALITY-HD. The research productivity of the department in which the entry level economist did their Ph.D. work is measured by QUALITY-PD (producing department), while QUALITY-HD (hiring department) is given by the productivity of the department in which the entry level economist accepts his or her first job. QUALITY-PD can be interpreted as a human capital variable reflecting the hiring department's anticipation of relatively high expected marginal productivity in research. Alternatively, it can be interpreted as an indicator of "prestige" which Youn |1988~ argues is a key factor in the operation of academic labor markets. In either case, QUALITY-PD is expected to have a positive and significant effect on entry level salaries.

The quality of the hiring department can have an effect on beginning salaries separate and apart from QUALITY-PD. The first thing to note about QUALITY-HD is that large numbers of entry level economists accept positions in departments that do not have faculty that publish in top journals. We refer to these non-publishing departments as "unranked." In contrast, the 243 departments that appear in Hirsch et al.'s |1984~ list are referred to as "ranked." In our survey 93 of the 268 new entrants in the sample took positions in unranked departments and 175 were hired by ranked departments. The second thing to note about QUALITY-HD is that within the subsample of those taking jobs in ranked departments we expect competition to lead the highest quality departments to bid for and hire the most talented new entrants. Thus, QUALITY-HD is expected to positively influence entry level salaries.(10)

The field of specialization of each entry level economist reported by the hiring department was used to construct a set of dummy variables for Journal of Economic Literature specializations which are denoted as FIELD. As indicated in Table I, the field of specialization with the mean salary closest to the overall mean is used as the dummy variable base in estimating the model. The cell sizes for individual specializations (microeconomics, labor economics, etc.) are not large enough to make reporting of individual coefficients meaningful. Therefore, we consider only whether the field variables are jointly significant and do not report coefficients for individual specializations. All other variables in Table I are interpreted in a straightforward manner.

Model Specification

Several specifications were explored and the effects of being hired into ranked and unranked departments were determined to be highly significant. We first estimated a pooled equation and investigated whether separate earnings equations should be specified for the those hired into ranked and unranked departments. In this pooled equation a dummy variable was used to distinguish those hired by ranked and unranked departments and a Chow test for interactions between ranked and other categorical variables (p-value = 0.0001) indicates that the determinants of earnings are significantly different at the 5 percent level. Therefore, estimating separate earnings equations for ranked and unranked departments yields superior overall explanatory power in our analysis of entry level economists salaries.

Equations 2.1, 2.2, 2.3 and 2.4 of Table II present our estimates of equation 1. Equation 2.1 shows the basic model for those hired by ranked departments and equation 2.2 shows the comparable estimate for unranked departments. Equations 2.3 and 2.4 expand upon the basic model by including the variable FIELD. In interpreting the results we focus first on sex and age and then briefly consider the effects of other variables.

The Effects of Gender and Age

As explained by Raymond, Sesnowitz and Williams |1988~, the coefficient on the dummy gender variable in equation 2 can be interpreted as a measure of labor market discrimination when all other relevant determinants of earnings have been included in the estimating equation. Similarly, the coefficient on the age variable measures the market treatment of newly hired economists with differing age levels.

The empirical results in Table II are quite clear; the age and gender variables are robustly insignificant. The results for the age variable may be as expected, but for females they may be quite surprising, at least to some observers. There are two contrasting reasons why gender could be a significant determinant of earnings. First, there is an established body of empirical evidence, mostly from the 1970s, relating to male and female academic salaries, and the predominant finding is that females earn significantly less at entry and the differential widens across time.(11) A second reason is that some observers believe that in the 1980s affirmative action programs escalated the competition for female faculty creating the possibility of a female premium.(12) Two recent studies for the 1980s reach contradictory conclusions. Rickman |1984~, studying a relatively small, teaching-oriented state university in the 1982-83 academic year, finds differentials quite consistent with the studies of the 1970s. In contrast, Raymond, Sesnowitz and Williams |1988~ investigate a large, comprehensive state university and find no measured discrimination in the 1983-84 academic year. In fact, Raymond, Sesnowitz and Williams |1988, 48~ report reverse discrimination when rank variables are added to their basic estimating equation, which, unlike most other studies, includes a personal measure of productivity (published research). However, the finding of reverse discrimination is not emphasized and, on balance, the absence of significant gender related differentials is stressed.

The insignificant results for FEMALE and AGE in Table II were further investigated by taking interactions with other variables into account. Table III reports Chow tests for the joint significance of particular variables of interest and their interactions with other variables. Large p-values (ranging from 0.461 to 0.943) indicate the absence of significant interaction TABULAR DATA OMITTED between FEMALE and other variables. This confirms that the nominal salary differential found in the raw sample data and reported in the appendix is entirely accounted for by the differences in male and female values of other variables in the earnings equation. Thus, our empirical results diverge from the numerous studies of male-female academic salary differentials in the decade of the 1970s and provide no support for the view that affirmative action has escalated the competition for females and created a female premium. In contrast, our results for FEMALE are consistent with Raymond, Sesnowitz and Williams' |1988~ finding of an absence of a gender related salary differential. While their results are institution specific, ours apply across the entire entry level market for economists.

Chow Tests

 Ranked Departments Unranked Departments

 with without with without
 intercept intercept intercept intercept

FEMALE 0.9432 0.9249 0.5343 0.4607

AGE 0.8161 0.8086 0.0824 0.0817

QUALITY-PD 0.2289 0.2061 0.7785 0.9348

QUALITY-HD 0.0078 0.9666 -- --

*p-value (significance level).

The results in Table III for AGE indicate that there are no significant interactions in ranked departments; thus neither gender nor age matters in this segment of the entry level market. However, the Chow test indicates that AGE and its interactions are significant at just below the 10 percent level in unranked departments. This suggests that age in combination with other variables does appear to be of some marginal consequence in unranked departments.

The Influence of Other Variables

For new hires in ranked departments TERMINAL DEGREE, HIGHEST DEGREE, PUBLIC, BUSINESS SCHOOL and INCOME are significant and have positive signs indicating that finishing one's Ph.D., taking a job in a Ph.D. granting department, in public college or university, in a business school and at a location with a relatively high cost of living all contribute to higher entry level salaries. QUALITY-HD is also positive and highly significant implying that the greater the research productivity in the hiring department, the larger the entry level salary. Finally, in the sub-sample of economists hired into ranked departments, FIELD is jointly significant.(13) NEW HIRES and, surprisingly, QUALITY-PD are not significant.

The results for the unranked departments are similar in some ways but quite different in others. As with ranked departments, TERMINAL DEGREE is highly significant. BUSINESS SCHOOL is also significant but, importantly, the size of the coefficient is much larger and the level of significance much lower than in ranked departments. In contrast to ranked departments, NEW HIRES is significant and HIGHEST DEGREE, PUBLIC, and INCOME are all insignificant in unranked departments. A further difference is found in the effects of specialization; FIELD is jointly insignificant in unranked departments. Finally, QUALITY-PD is marginally significant at the 10 percent level in equation 2.2 but is insignificant in equation 2.4 when FIELD is included.

The results for the quality variables warrant further comment. We explored alternative measures of quality using conventional departmental rankings based on research productivity, e.g., top-twenty, top-thirty etc. In this approach QUALITY-PD and QUALITY-HD were dummy variables distinguishing between highly ranked departments and all others. Of course, the dividing line between top departments is somewhat arbitrary. Experimentation with alternative definitions of top departments revealed that there is no unique dividing point that distinguishes quality departments from all others. For this reason we used the continuous quality measures on which the rankings are based.(14) Finally, in a fashion analogous to our analysis of age and gender and their interactions with other variables we investigated the joint significance of the quality variables. Table III reports the results of the Chow tests. Looked at from this perspective, only the intercept term of QUALITY-HD matters in ranked departments and QUALITY-PD is not jointly important in ranked or unranked departments.

Estimates of Salary Premiums

To analyze the effects of the categorical variables on entry level salaries we apply Kennedy's |1981~ dummy variable procedure to the earnings equations in Table II. Table IV reports exact percentage impacts of the categorical variables which are significant at the 10 percent level or higher. Column 1 shows the marginal impacts of the significant determinants of the entry level salaries of 175 economists hired by ranked departments. Column 2 shows similar information for the 93 economists hired by unranked departments. For those hired by ranked departments, completing the Ph.D. degree is associated with a 4.4 percent salary premium; if the position is in a publicly supported university there is a 3.5 percent premium; business schools pay a premium of 2.1 percent; and Ph.D. granting departments are associated with a 3.9 percent premium. For those hired by unranked departments the only significant dummy variables are TERMINAL DEGREE and BUSINESS SCHOOL, which have premiums of 7.7 and 11.3 percent, respectively.

Exact Percentage Salary Premiums for Categorical Variables(*)

 Ranked Departments Unranked Departments






* Calculated using Kennedy's |1981~ procedure.

NS--Variable is not significant.



This study uses a unique data set to investigate the effects of gender and age on entry level salaries of economists hired in the 1987-1988 academic year. The results are clear and robust; gender does not matter in the entry level market for academic economists. Age does appear to matter, but only in non-research oriented departments and only in terms of its interactions with other determinants of earnings. The analysis identifies a number of other factors that significantly influence starting salaries. Among economists hired by ranked departments, completing the terminal degree, field of specialization, the cost of living in the county where the job is located, the quality of the hiring department, accepting a position in a publicly supported college or university, a department that offers a Ph.D. degree and a department located administratively in a business school all significantly and positively influence the entry level salary. The determinants of entry level salaries of those hired by unranked departments include completion of the Ph.D. degree, the number of entry level economists hired by the department and taking a position that is administratively located in a business school.

JOHN P. FORMBY, WILLIAM D. GUNTHER, and RYOICHI SAKANO The authors are respectively, Professor of Economics, Professor of Economics, University of Alabama, and Assistant Professor, North Carolina Agricultural and Technical State University. The authors thank Donald Frey, Barry Hirsch, James Lindley, James Marchand, Terry Seaks, Gary Zarkin, three anonymous referees and the editors for helpful comments on an earlier version of this paper. The usual caveats apply.

1. The 469 departments were selected in a two step procedure. First, all U.S. departments included in Hirsch et al.'s |1984~ list of 243 ranked departments were included. Second, all U.S. departments not in this list, but which sponsored a chapter of Omicron Delta Epsilon, the International Honor Society in Economics were included.

2. Individual entry level economists are the unit of observation, but the hiring departments are the respondents to the survey. A number of the 258 responding departments reported multiple hires in 1987-88 and some reported no hires resulting in a total of 268 observations (new hires).

3. For a summary of the response rate see Report of AEA Committee on the Status Women in the Economics Profession |1990~.

4. Named after Professor Ernst Stromsdorfer of Washington State University who conducts the survey and tabulates the results.

5. The AEA Chairperson's Group is an informal organization that holds a single breakfast meeting at the AEA convention each year. Its purpose is to exchange information and discuss common problems.

6. Other valuable information that would be useful in estimating earnings equations is collected. For example, the Stromsdorfer Survey contains detailed questions relating to departmental summer research support, housing supplements, moving costs and commitments to provide microcomputers to newly hired entry level economists. However, the response rate on the fringe benefit questions is substantially below the rate for other questions. Further, except for the field of specialization and producing department ("institution of origin"), the Stromsdorfer Survey does not collect information on individual new hires.

7. Tables summarizing descriptive statistics, the responding departments, the universities at which the 268 economists in our sample did their graduate work and the identity of the departments included in our study are available on request.

8. As an alternative formulation to equation 1, the logarithm of real earnings could be used as the dependent variable. This would involve dividing nominal earnings by the cost-of-living index. However, we estimate equation 1 rather than the alternative because we know that the cost of living proxy we are using is imperfect and contains measurement error. As a consequence, the estimated coefficient of the cost of living proxy variable is biased downward toward zero and the estimated coefficients of other variables could also be biased. This follows from the well-known 'errors-in-variables' or 'unobservable variables' problem in econometrics. But since the cost of living is expected to be relevant to the determination of entry level salaries we include the proxy in our estimation. We note that the more random noise the proxy variable contains, the greater the downward bias in its estimated coefficient. Thus, we expect the estimated coefficient of the per-capita-income variable to be between zero and one, which it is in each of our estimates of equation 1.

9. The quality of economics departments has been measured in a number of studies and faculty research productivity is typically the key or only ingredient. Hogan |1973~, Niemi |1975~, Graves, Marchand and Thompson |1982~ and Hirsch, Austin, Brooks and Moore |1984~ all assume quality is a function of faculty research productivity. The latter three studies use the same methodology and measure quality in terms of the total number of pages published in twenty-four leading journals standardized to American Economic Review equivalent pages. We use the data reported in Hirsch et al., which is the most recent.

10. One can conceive of an inverse relationship between QUALITY-HD and entry level salaries. This could be caused by higher quality departments offering new faculty members greater opportunities for learning from and collaborating with distinguished scholars, enhanced mobility following a short appointment in the department, or non-pecuniary advantages associated with a position in a high prestige department. A lower starting salary, thus, could be a negative compensating differential which is balanced against the greater expected future income stream or psychic benefits. But, we expect that competition among research oriented departments will dominate and that QUALITY-HD will have a positive influence on entry level salaries.

11. For a summary and citations to the relevant literature see Jusenius and Scheffler |1981~, Hirsch and Leppel |1982~ and Jackson and Lindley |1989~.

12. This point was suggested by an anonymous referee. The same point was raised by seminar participants at East Carolina and Wake Forest Universities where an earlier version of the paper was presented.

13. As noted above, the cell sizes for individual specializations are not large enough to make reporting of individual coefficients meaningful. Nevertheless, it is informative to reveal somewhat more about the effects of FIELD. We used the classification codes prevailing at the time of the survey, and the Journal of Economic Literature's (JEL) Field 300 (monetary and fiscal theory and institutions) was the closest to the mean and was used as the base against which other fields were compared. Using equation 2.3 for ranked departments the largest premium went to JEL Field 700 (Natural Resources), which was 6.7 percent above Field 300. There were two Fields at the other extreme, Field 010 (General Economics) received a 4.1 percent discount and Field 100 (Economic Development) was associated with a 4.3 percent discount.

14. The experimentation proceeded by first defining quality in terms of top-twenty, then top-thirty, then top-forty, etc. and seeking to identify a dividing line where this measure of quality was significant for top departments but not for other departments. In the case of QUALITY-PD there was no indication of such a dividing line once the sample was segmented into those hired by ranked and unranked departments. For QUALITY-HD the experimentation suggested a dividing line around the top-eighty to top-one-hundred departments. Using a Chow test we find a statistically significant break point at rank eighty. We also find such a break point at the rank one-hundred. However, the research productivity of departments ranked between eighty and one-hundred are close and the variable is continuous. Analyzing departments in increments of twenty is somewhat arbitrary and many break points between eighty and one-hundred were also found. Thus, using dummy variables to measure quality involves a subjective judgment concerning the dividing line between departments. For this reason we used the continuous values of research productivity as our measure of quality.


American Economic Association on the Status of Women in the Economics Profession. "Report." American Economic Review, May 1990, 486-89.

Graves, Phillip E., J. Marchand, and Randall Thompson. "Economics Departments Rankings: Research Incentives, Constraints and Efficiencies." American Economic Review, December 1982, 1131-41.

Hirsch, Barry T., and Karen Leppel. "Sex Discrimination in Faculty Salaries: Evidence from a Historically Women's University." American Economic Review, September 1982, 829-35.

Hirsch, Barry T., R. Austin, J. Brooks, and J. Bradley Moore. "Economics Department Rankings: Comment." American Economic Review, September 1984, 822-26.

Hogan, Timothy D. "Rankings of Ph.D Programs in Economics and the Relative Publishing Performance of Their Ph.D.'s: The Experience of the 1960's." Western Economic Journal, September 1973, 429-50.

Jackson, John D., and James T. Lindley. "Measuring the Extent of Wage Discrimination: A Statistical Test and A Caveat." Applied Economics, April 1989, 515-40.

Johnson, George, and Frank P. Stafford. "Lifetime Earnings in a Professional Labor Market: Academic Economists." Journal of Political Economy, May/June 1974, 549-69.

Jusenius, Carol L., and Richard M. Scheffler. "Earnings Differentials Among Academic Economists: Empirical Evidence on Race and Sex." Journal of Economics and Business, Winter 1981, 88-96.

Katz, David. "Faculty Salaries, Rates of Promotion and Productivity at a Larger University." American Economic Review, June 1973, 469-77.

Kennedy, Peter E. "Estimation With Correctly Interpreted Dummy Variables in Semilogarithmic Equations." American Economic Review, September 1981, 801.

Niemi Jr., Albert W. "Journal Publication Performance during 1970-1974: The Relative Output of Southern Economic Departments." Southern Economic Journal, July 1975, 97-106.

Raymond, Richard D., Michael L. Sesnowitz, and Donald R. Williams. "Does Sex Still Matter? New Evidence from the 1980s." Economic Inquiry, January 1988, 43-58.

Rickman, Bill D. "Does Sex Matter?" Quarterly Journal of Business and Economics, Spring 1984, 47-57.

Youn, Ted I. K. "Studies of Academic Markets and Careers: An Historical Review," in Academic Labor Markets and Careers, edited by D. Brenneman and T. Youn. Philadelphia: Falmer Press, 1988, 8-27.
COPYRIGHT 1993 Western Economic Association International
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1993 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Formby, John P.; Gunther, William D.; Sakano, Ryoichi
Publication:Economic Inquiry
Date:Jan 1, 1993
Previous Article:Professional achievements and gender differences among academic economists.
Next Article:Theory and experiments on spatial competition.

Terms of use | Copyright © 2016 Farlex, Inc. | Feedback | For webmasters