(June 2014) Regional economists use the shift-share analysis method to determine how much of regional job growth can be attributed to national trends and how much is due to unique regional factors. We adapted this method to examine growth in biomedical postdoctoral employment in U.S. research education institutions.
We looked at employment growth in three time periods corresponding to shifts in the National Institutes of Health funding levels. From October 1998 to September 2003, the NIH budget doubled, growing at about 15 percent annually. From 2003 to 2009, growth in the NIH annual budget slowed to between 1 percent and 3 percent per year, growing by about 8 percent. National growth in biomedical postdoctoral employment rose from 14.6 percent to 19.6 percent, then fell back to 15.2 percent in the respective periods.
To conduct shift-share analysis, we split biomedical postdoctoral employment job growth into three components: subfield mix effect, national postdoctoral employment growth effect, and demographic subgroup competitive advantage. For the purposes of this analysis, we focus primarily on each demographic subgroup's competitive advantage.
- The Subfield Mix Effect represents the demographic subgroup's share of biomedical postdoctoral employment growth that is explained by biomedical postdoctoral employment growth of the specific subfield at the national level.
- The National Growth Effect explains how much of the demographic subgroup's growth in biomedical postdoctoral employment may be explained by the overall growth of national biomedical postdoctoral employment: If employment of biomedical postdoctoral trainees in the nation as a whole is growing, one would generally expect to see some positive change in each subfield for each demographic subgroup.
- The Demographic Subgroup's Competitive Advantage in a subfield indicates how much of the change in a given subfield is due to some unique (unmeasured) advantage that the subgroup possesses because the growth in biomedical postdoctoral employment cannot be explained by expected change. Here expected change is the sum of the subfield mix effect and the national growth effect. A positive competitive effect for a demographic subgroup in a subfield indicates the subgroup is outperforming national trends (both overall national trends and national trends in that specific industry). A negative effect means that subgroup in a subfield is underperforming compared to national trends.
By definition, the sum of the shift-share components for any demographic subgroup must equal the total change in biomedical postdoctoral employment for that same subgroup over all subfields in the time period. Also, the sum of all demographic subgroups’ competitive advantages across the subfields must equal zero.
No clear story emerges about how competitive advantage in postdoctoral biomedical employment differs by citizenship/visa status. Both male and female “foreign-born” (temporary visa holders) experienced better-than-expected growth in biomedical postdoctoral employment during the rapid growth in funding from 1998 to 2003.
Their competitive advantage in this period was greater than in 1993-1998. Male and female “U.S.-born” (U.S. citizens/permanent visa holders) competitive advantage in the 1998-2003 period was lower than in the 1998-2003 period. However, with the slowing growth of funding in 2003-2009, U.S.-born men and women saw increases in their competitive advantage, while foreign-born men and women saw substantial decreases in competitive advantage.
One possible explanation of the shifts in competitive advantage observed over these three time periods may be that NIH funding growth in biomedical research opened up better job opportunities (faculty positions in research institutions, well-paying nonfaculty positions, or nonresearch positions) for citizens and permanent visa holders, leaving a less competitive U.S.-born pool vying with foreign-born applicants for postdoctoral positions. When biomedical research funds decreased again, more competitive U.S.-born applicants remain in the pool seeking postdoctoral appointments.
Another explanation would be that the relative quality of the pool of applicants from each demographic subgroup changed over time, with an improvement in foreign-born applicants between the 1993-1998 and 1998-2003 periods and a decline in quality of these applicants between 1998-2003 and 2003-2009. This second explanation would be consistent with shifts in pull factors prior to 2009 when effects of immigration policy and booming global real estate markets might have affected foreign-student graduate enrollments in the United States as well as foreign workers seeking employment in faculty positions.
The effect of subfield mix on biomedical postdoctoral employment growth is clearer. The size of the subfield mix effect has increased with each successive period, regardless of citizenship/visa status and gender. For all demographic subgroups, the effect of their distribution across subfields has become increasingly important over time. Growth of postdoctoral employment in new interdisciplinary fields such as biomedical engineering and ecology remains strong, but growth in traditional life science fields such as anatomy and zoology continues to slow or decline. Additional graphs present growth in each subfield as a proportion of the number employed in biomedical postdoctoral positions at the beginning of the period (1993-1998, 1998-2003, and 2003-2009).
The general expectation based on national growth in postdoctoral appointments and the pattern of growth in subfields would be that all demographic subgroups would show increases in biomedical postdoctoral employment in this period. The distribution of the employment growth across the subfields negatively affected biomedical postdoctoral employment growth among male citizens/permanent visa holders (U.S.-born) and female temporary visa holders (foreign-born). Only postdoctoral employment among male citizens/permanent visa holders did not increase to the extent anticipated, as reflected in their negative competitive advantage for the period.
All demographic subgroups should have experienced substantial biomedical postdoctoral employment growth, based on the national growth effect plus the subfield mix effect. However, competitive disadvantage reduced growth in postdoctoral employment to nearly half the expected growth for male and female citizens and permanent visa holders.
National growth in biomedical postdoctoral employment did not slow dramatically. Differential growth of postdoctoral employment across biomedical subfields affected all demographic subgroups negatively. Biomedical postdoctoral employment grew less than expected for all subgroups except for female citizens/permanent visa holders—the only group with a positive competitive advantage in this period.
Notes on the Data and Definitions
The data source for the table and graph used in this analysis is the Graduate Students and Postdoctorates in Science and Engineering (GSS) from 1993 to 2009. This annual survey of all academic institutions in the United States granting research-based master’s degrees or doctorates in science, engineering, or selected health (SEH) fields provides data on the number and characteristics of graduate students, postdoctoral (postdoc) appointees, and doctorate-holding nonfaculty researchers in these fields.
Definitions used in this analysis follow the NSF-NIH classification of fields included in the biomedical, behavioral and social, and clinical sciences. See NIH Data Book http://report.nih.gov/nihdatabook/.
Definitions of demographic subgroups similarly rely on classifications consistently available in the GSS from 1993 through 2009, with U.S.-born being U.S. citizens or permanent visa holders and foreign-born being temporary visa holders.
In pdf documents showing growth components as a percent of initial number of postdoctoral appointees (1993-1998, 1998-2003, and 2003-2009), values less than 0 indicate a proportionate decrease. Negative values cannot fall below -100. Values greater than 0 indicate a proportionate increase: 100 represents a doubling in the subfield, and values over 100 more than a doubling.
Excel file with raw data and notes.