Is It All Worth It? The Experiences of New PhDs on the Job Market, 2007–10

Abstract

The authors describe job market experiences of new PhD economists, 2007–10. Using information from PhD programs’ job candidate Web sites and original surveys, they present information about job candidates’ characteristics, preferences, and expectations; how job candidates fared at each stage of the market; and predictors of outcomes at each stage. Some information in this article updates findings of prior studies. However, design features of the data used in this article may result in findings that are more generalizable. This article is unique in comparing premarket expectations and preferences with post-market outcomes on the new PhD job market. It shows that outcomes tend to align with premarket preferences, and candidates’ expectations are somewhat predictive of their outcomes. Several analyses also shed light on subgroup differences.

Keywords:

• economics PhD
• job market
• preferences
• salary

JEL codes:

• A2
• J4

ACKNOWLEDGEMENTS

The authors gratefully acknowledge support from the Sloan Foundation's Scientific & Engineering Workforce Program, the Population Studies Center's Eva L. Mueller New Directions in Economics and Demography Fund, and the A. Regula Herzog Young Investigators’ Fund. They are thankful for feedback from Charlie Brown, David Lam, and Miles Kimball about the Job Seekers Project and its surveys, and thank Martha Bailey, Trivellore Raghunathan, Matthew Shapiro, and Jeffrey Smith for comments and feedback about this article.

Notes

The structure of the market is described in detail in annual job market guides by John Cawley (see, e.g., Cawley 2014).

The Job Candidates Web site of the NBER is accessible at http://www.nber.org/candidates/.

For example, our numbers line up favorably with Coles and colleagues (2010, 191–92), who recorded 978 to 1,022 individuals who used the AEA signaling mechanism between 2007 and 2009. Their surveys show that about two-thirds of current PhD students on the job market (similar to the sample frame in this article) reported signaling and that about two-thirds of those who signaled were students. Thus, if about 1,000 per year signaled, then 667 were students. These 667 were approximately two-thirds of the number on the market, so the total market is around 1,000.

Excluding non-U.S. institutions holds the sample size virtually unchanged from cohort to cohort. Sample sizes are 832 in 2007–8, 845 in 2008–9, and 826 in 2009–10. Excluding “unranked” institutions makes the samples yet more similar. Throughout the article, we report results using all possible data.

Program rank comes from the U.S. News and World Report (2009) rankings of graduate programs in economics.

Identified from photographs or using the Baby Name Guesser at http://www.gpeters.com/names/baby-names.php to assign the most likely gender based on first name.

Research fields are hand-coded into 28 categories. The greatest number of fields coded for any one candidate was eight. In the coded data, 91 percent of candidates had four or fewer fields.

The 2007–8 and 2009–10 post-market surveys were fielded beginning in August. The project fielded the 2008–9 post-market survey in November rather than August.

Variables used in creating all or some of the weights include gender, PhD program rank, region of undergraduate education, age, additional degrees, teaching experience, number of journal publications, undergraduate field, whether CV was posted on Web site, country of PhD institution, field, and citizenship.

Results from the surveys indicate that location of undergraduate institution is a good proxy for citizenship. We identify Asian countries following the classification scheme of the Population Reference Bureau (2008).

The only qualitative difference when we exclude non-U.S. PhDs is that a greater proportion of the U.S. PhD sample is from U.S. undergraduate institutions, and slightly more in each ranking level of institutions. The latter is mechanical because rankings are based on U.S. programs only, and the former is intuitive. Other than these differences, all estimates are within one percentage point of those presented in the table.

Most (66 percent) job candidates from unranked departments were from lower-ranked economics departments in the United States. A sizable minority (30 percent) were from departments outside the United States, and a small number (4 percent) were from departments in fields closely related to economics, such as business or public policy.

The job type preference variable was not identical from year to year. Specifically, in the 2007–8 premarket survey, we asked about preferences over job setting (university, four-year college, postdoctoral fellowship, nonacademic research, and nonacademic nonresearch), while in later years we asked about job types more specifically (assistant professor at four-year college, assistant professor at university, postdoctoral fellow, nonacademic researcher [e.g., researcher at a think tank, government research unit, central bank, or international financial organization], private sector researcher). However, the descriptions were similar, and a chi-square test of distribution in most-preferred job types between cohorts does not reject the null hypothesis that the distributions are identical (p-value.45).

The Pearson chi-squared test statistic is 30.34 (36 df), p = 0.734. The correlation coefficient is −0.0074. The coefficient from an OLS regression of expected rank on rank of PhD institution is 0.009 (SE 0.09).

Regression conducted using survey weights for response to both surveys. N = 247.

It is possible that the job seekers who had time to participate in a survey during the AEA meetings had less busy interview schedules than other job seekers.

Excluding non-U.S. PhD programs, significance patterns and signs from table 5 are all the same, with one exception: Courses as primary instructor became statistically significant at the 5-percent level in column 3. The coefficient is 1.83 (SE 0.91).

Despite higher yields from fly-outs to job offers, the average number of job offers to job candidates with Asian undergraduate degrees (2.5 job offers, SE 0.2) remains below that of those with degrees from other countries (3.2 job offers, SE 0.1). Given that job candidates from Asian undergraduate institutions apply to more jobs on average (110.6 applications, SE 5.3) than those from other locations (106.7 applications, SE 2.4), these job candidates have lower success rates than others between the application and fly-out stage. Despite higher yields from fly-outs to job offers, the average number of job offers to job candidates from PhD programs ranked 21 to 50 (2.5 job offers, SE 0.1) remains below that of those from the top 20 programs (3.9 job offers, SE 0.2). Given that job candidates from lower-ranking programs apply to more jobs on average (120 applications, SE 4.5) than those from higher-ranking programs (97.6 applications, SE 3.6), these job candidates have lower success rates than others between the application and fly-out stage.

For example, if the job postings in a given year disproportionately seek health economists, health economists are likely to respond by applying to a large number of jobs. Then, because demand for their skills is high, the applications of health economists may also be more likely to be successful.

We have included visiting assistant professor and lecturer positions at universities in this category. The “tenure track” statistics in panel C provide some insight into this.

We have included visiting assistant professor and lecturer positions at colleges in this category. The “tenure track” statistics in panel C provide some insight into the fraction who are on the tenure track. Stock and Siegfried (2001) found that 6 percent of new PhDs about whom they received information about the 1996–97 market were employed in full-time permanent U.S. positions at Carnegie BA/BS institutions. This is close to our estimate of 8 percent. It should be noted that they include only “full-time permanent” positions, which excludes adjunct, lecturer, and visiting positions, whereas we include these positions here if the candidate so classified him/herself.

Includes positions with consulting firms.

Restricting data to the candidates from U.S. PhD institutions only, the only notable difference is that in table 7 a slightly lower percentage of job candidates prefer postdocs (3 percent versus 5 percent in the full sample), while 9 percent prefer colleges, and 63 percent prefer universities. One possible explanation for this is that for foreigners who want to get into the U.S. job market, a postdoctoral fellowship is one way to make connections and improve English skills.

Indeed, more than two-thirds of job candidates in 2008–9 and 2009–10 reported that “a high personal income” was either “not important,” “a little important,” or “moderately important.” Only a quarter rated high income as being “very important,” and 7 percent rated it “extremely important.”

Calculated using the Bureau of Labor Statistics’ CPI-U for June 2001 and June 2010 (Crawford and Church 2014).

All quartiles are virtually identical when excluding those from non-U.S. PhD institutions. Means are $1,500 to $2,000 higher when restricting to U.S. PhDs. However, even the overall mean, which is $2,000 in difference and has the largest sample size, is not statistically significantly larger than that of the main sample (p-value = 0.12).

Results do change when excluding non-U.S. PhDs. In this discussion, we only discuss results that changed the takeaways from this table. In column 1, the estimate for “Preferred job type: NGQO” dropped to 4.68 (SE 2.89, significant at the 1-percent level). That is, for candidates who preferred a job at an NGQO, they were about five times more likely to become employed at a college than a university. This is half the effect size that is presented in the table, but in the same direction. Additionally, the relative risk ratio for unranked PhD programs increased to 3.48 (SE 2.14), becoming statistically significant at the 1-percent level. Column 2 was virtually identical. In column 3, the estimate on “Preferred job type: NGQO” more than doubled, to 117, and the coefficient on “Preferred job: Postdoc” tripled (to 38.45, SE 62.9) and became statistically significant at the 5-percent level. These findings lend more support to our finding that openness to jobs outside traditional academia greatly increases the chances of accepting a nonacademic job. In column 4, the estimate on “Preferred job type: NGQO” changed to almost zero and is statistically significant, indicating that U.S. PhDs who prefer NGQO jobs virtually never receive postdoctoral fellowships. The estimate on “Preferred job: Postdoc” also increased fivefold (to 105.9, SE 175.11), indicating that a U.S. PhD who prefers a postdoc is many, many times more likely to secure a postdoc over a university job. This is statistically significant at the 0.1-percent level. The estimates on lower-ranking PhD programs also increased. Additionally, having three or more journal publications was associated with close to no chance of receiving a postdoc over a university position (0.00, SE 0.00).

The coefficient on “Undergraduate location: Other” decreased in magnitude, to −4.64 (SE 2.59) and lost its statistical significance when using only the U.S. PhD sample. There are no other changes in significance level, and all statistically significant coefficients are within 10 percent of the value shown in table 11.

Excluding the non-U.S. PhDs, the coefficients on “PhD program rank: 21–30” and “Three or more journal publications” both declined by less than 10 percent, and lost their marginal statistical significance.