Summary: EMSI has developed a new index to help regions better understand employment skills gaps. This methodology is an attempt to provide a straightforward but comprehensive technique for sensing shortages of workers so that regional planners can respond where needed. Of the 41 occupations included in our analysis, 30 have a lower-than-average supply of potential workers. The shortage is pronounced among occupations in the healthcare, education, business & finance, and architecture & engineering sectors.
In our previous post, we discussed occupations with potential skills shortages, and observed that jobs that have a lot of demand show increases in both employment and earnings over time. The simple progression is well known: employers target skilled workers that will add value to the company, these same workers are often recruited by other companies, and the resultant competition creates higher wages and benefits. This phenomenon is clearly seen in areas such as software development, where competition between tech companies for talented programmers drives up wages and benefits.
In this post, we want to elaborate on the supply side of this equation. Do occupations with greater demand also experience a commensurate increase in qualified workers? And how do we accurately measure the supply side of this equation?
To lay this out, first imagine that a company is trying to hire an accountant. It will post a job opening on its website, on job search websites, in the local newspaper, at the nearest university, and with local employment agencies. Because of the different places the job was posted, a variety of people will see the new opening. Some of these will be active jobseekers (those who have already submitted a resume for a specific position) while others will be passive (those who have posted their resumes online for employers to peruse). To better understand the different supply of workers, we can place each type of jobseeker into one of three categories: (1) new graduates, (2) unemployed workers with the requisite skills to take the position, and (3) compatible workers currently employed elsewhere who have the requisite skills and the proper incentives to change jobs.
Here is a detailed description of each category and where the data comes from:
This is the most commonly used measurement of supply. The primary data source is the Integrated Postsecondary Educational System (IPEDS). IPEDS records the number and type of graduates from every postsecondary institution in the nation that accepts Title IV funding (e.g., Pell Grants, Stafford Loans, etc.). As with any data source, it has certain weaknesses. It can sometimes fall short in terms of accuracy. In some program areas, there can be duplication of graduates. In others, the figures may over-estimate the number of workers who become certified for employment. Additionally, this source does not include the marginal number of apprenticeship completions that also align to specific occupations. However, it is the best source to look to for national data on graduates.
National unemployment data is collected by the US Department of Labor through the Current Population Statistics Survey (CPS). In addition to providing aggregate national unemployment figures, the CPS divides this data into occupations according to Standard Occupational Classifications (SOC). This allows us to tell the difference between occupations in which the skilled labor is being nearly maxed out and those where much of the skilled labor is not being utilized. For example, in 2010 the unemployment rate for occupational therapists was a strikingly low 1% compared to 34.5% for telemarketers. The higher the proportion of skilled labor that is employed in that position, the fewer workers can be assumed to be available on the supply side.
This is the trickiest and most innovative measurement involved in the analysis. The purpose of this measurement is to estimate the number of workers who are willing and able to switch occupations into categories where their skills are more highly valued than in their current positions. In order to meet the “willing and able” criteria, these workers must a) possess compatible knowledge, skills, and abilities, and b) have the proper wage incentive to consider making a career change. Compatibility is measured in terms of EMSI’s proprietary compatibility index, which is based on O*NET competency calculations. (This is a complex process that involves finding the closest match between SOC codes in terms of both ability and importance.)
These categories help us perform a gap analysis, which provides an objective, data-driven look at the pipeline of skilled workers. For each group, this is generally what we are looking for:
- Regional educational institutions (colleges, universities, or other institutions that report to IPEDs) graduating (or reporting) enough students to match the openings for a particular occupation or set of occupations;
- Qualified unemployed workers in the region;
- Enough compatible workers who could transition from other jobs.
A “gap” occurs if the sum of workers from each category is below a certain threshold when compared to the amount of openings for a particular occupation.
When Supply > Demand
If we find an adequate pipeline of workers, we then look at things like wages and other potential opportunity costs, which might be keeping qualified workers from taking the jobs. Measuring occupational wages is pretty straightforward and involves comparing current wages to the median wage offered in another occupation. If the employer is not willing to pay enough for a certain worker, it could explain why there is a shortage of workers.
If there are qualified candidates in the workforce but employers cannot find them, we need to better understand why. What combination of factors is required to draw workers from their current jobs (or regions) toward new jobs (or new regions)? There are a host of other opportunity costs associated with making a career change. If the cost exceeds the perceived benefits of the change, the individual will likely not switch. Here are some typical examples:
- A person feels they have already invested time and money in their current position and a risk wouldn’t necessarily reap a greater reward;
- They think that their current wages will increase if they just stick it out;
- They are constrained to their current job and/or city for family reasons, or they are drawing a second paycheck through another employer;
- They are unable to move (e.g., can’t sell the house); or
- They feel reluctant to leave their current place of employment because of the recession.
At EMSI we have begun to work on a way to analyze this. Based on our compatibility index (see the “Compatible Workers” section), we assumed that for jobs with compatible skills and comparable wages, about 6.7% of the workforce would be willing to change and actively seek a new opening. To account for lower levels of compatibility, we decrease that percentage by one point for each drop in compatibility score. For instance, for jobs that are slightly less compatible but where a wage incentive still exists (such as computer software programmers and computer support specialists) the number is scaled down according to the level of compatibility.
Skills Shortage Analysis
The following table displays the 41 occupations that are commonly reported as having a shortage of skilled workers. The table includes the number of annual graduates, estimated unemployment, compatible workers in other similar occupations, a measure of the potential workers, and the average potential-workers-to-annual-openings ratio.
Lower numbers (those below the national average of 14 workers per job) mean that there are fewer potential workers to fill future vacancies, which is what we would define as an actual skills shortage. Conversely, higher numbers (those above 14 workers per job) mean there are more workers that could potentially fill job openings. Note: All of these occupations experienced substantial employment growth and earnings increases between 1999-2010. They are therefore considerably more likely to experience skills shortages than other types of occupations.
This data supports the idea that there is a shortage of skilled workers in 30 of the 41 occupations.
- In healthcare, education, business & finance, and architecture & engineering, 20 out of 28 listed occupations are lower than the national average benchmark of 14 potential workers per job.
- For occupations that fall below the line, some have low ratios due to an insufficient number of graduates (e.g., accountants & auditors and interpreters & translators).
- Some occupations have low ratios because they are so specialized that very few workers from other categories have the skills and necessary incentives to make the job transition (e.g., respiratory therapists, architects, and management analysts).
- For jobs that have a decent number of potential workers (e.g., above 14 jobseekers per opening), we would want to explore wages and learn more about the barriers preventing workers from taking those jobs.
- The occupations with the largest gaps are veterinarians, occupational therapists, accountants & auditors, pharmacists, and physician assistants.
- Occupations with the best potential pool of candidates are claims adjusters, physical therapy assistants, diagnostic medical sonographers, skin care specialists, and service unit operators for oil, gas and mining. Again, when there are larger pools of potential workers but lots of unfilled openings, we need to look at wages and opportunity costs.
1. Educational institutions with a particular focus on the skilled trades need to be looking closely at labor market data. They also need to have a very solid understanding of workforce demand. This should help facilitate smarter, more responsive education and training, which results in graduates equipped with skill sets that qualify for multiple careers in areas with plenty of demand for the foreseeable future. See this post on mechatronics for more on that.
2. For occupations where educational output is too low, policymakers can encourage greater enrollment and graduation. In-demand occupations requiring the most unique combination of skills should receive immediate attention from educators and workforce development stakeholders.
3. Educational institutions should also keep in mind that if a majority of vacancies can be filled with compatible workers who may or may not require further education, circumstances in the labor market might naturally draw compatible workers toward these positions. In such cases, greater educational output could result in oversupply and potentially unemployment for new graduates. An example of such a scenario could be operations research analysts; if wages continue to rise for this occupation, many people who are currently working as computer systems engineers and computer systems analysts would certainly consider submitting their resumes for operations research analyst openings.
4. Looking at the group of occupations above the 14.1 threshold can also provide meaningful insight. Among these occupations, there are a few textbook cases of educational overproduction — postsecondary administrators, massage therapists, and physical therapists assistants. If the postsecondary training programs associated with these occupations were looked at in isolation, we might be tempted to think that the demand and supply were relatively equal, but in fact each of these occupations is highly compatible with a number of other occupations, which are typically lower-paying.
5. Finding the optimal number of trained workers is complex, but it can be aided by making a quick study of various labor market data sources and measuring worker compatibility and wage incentives. This approach can yield a much more accurate answer than if we relied solely on traditional methods, such as comparing educational supply to projected occupational demand. The data presented in this analysis is by no means the final word on this topic but is hopefully a step in the right direction for more comprehensive analysis. We always encourage people to ground-truth the data in order to validate the findings.
Please contact us if you have any questions about this methodology or would like to prepare an analysis for your area.
Closing Note: As we do this, a topic that frequently comes up is the issue of licensure or certification. For example, many occupations do not require any formal certifications. If an insurance salesman wants to become a sales rep for wholesale or manufacturing, no additional postsecondary training is required. On the other hand, if a licensed practical nurse wants to become a registered nurse, it may take a year or more of additional training and new certification requirements. Given this, the amount of time required for workers to transition from one occupation can be difficult to determine. In our analysis we are simply trying to identify workers who are willing and able to make career transitions.
 There is a key difference between active and passive jobseekers. “Active” job seekers are people who have applied for a position, or who have submitted their resume for a specific position. “Passive” jobseekers are people who may have their resume posted online for potential employers to peruse, or they may occasionally search for new jobs. We are not seeking to quantify the number of passive jobseekers because this number would not be helpful for analysis.
 There is one important caution regarding these data. Unemployment by occupation data is recorded according to the most recent occupation in which a person was employed, which does not necessarily indicate the occupation in which they are most skilled or have the most experience. Nevertheless, the addition of this data source adds significantly to the robustness of this model. The source for the information used in this analysis was the Wall Street Journal’s infographic titled “Unemployment Rates: A Detailed Look.”http://online.wsj.com/article/SB10001424052748703791904576075652301620440.html.
 The genesis of these ideas came from two authors who have analyzed these questions in the past: Malcolm S. Cohen, in his book Labor Shortages as America Approaches the Twenty-First Century (Ann Arbor: University of Michigan Press, 1994); and Caroline Veneri’s article “Can occupational labor shortages be identified using available data?” Monthly Labor Review, 122, No. 3 (March, 1999). However, the author had added elements to the suggestions of both of these authors.
 The amount of literature on this topic is immense. Pertinent sources include the following: Assar Lindbeck and Dennis J. Snower. “Insiders vs. Outsiders.” Journal of Economic Perspectives, Volume 15, Number 1 (Winter 2001): 165–188. Dale T. Mortensen and Christopher A. Pissarides. “Job Creation and Job Destruction in the Theory of Unemployment.” Review of Economic Studies, 61 (1994): 397-415.
 The 6.7% rate was determined based on several recent studies that indicate between 12% and 19% of the American workforce in 2010 and 2011 is actively looking or a new job. (Jobvite 2010 Survey: http://web.jobvite.com/rs/jobvite/images/Jobvite_Survey_Jobseeker_Nation_2010_final.pdf; Delloite 2011 Survey: http://www.tlnt.com/2011/03/08/survey-74-of-workers-are-passive-job-seekers-ready-to-consider-a-move/.) Though these surveys do not deal directly with the types of career transitions that workers will make, it can safely be assumed that if a worker is considering making a career transition, he/she is most likely to transfer into a position with which they have some existing skill compatibility. The highest possible rate between two occupations in the EMSI model is 12.5%, but that would require near-perfect job compatibility, which does not exist for any two occupations. By default, the highest possible job transfer rate is 6.7% at a compatibility score of 98.
 The average applicants to vacancy rate recorded in this table is 14.1. This is based on research of employer search methods which was conducted including companies of various sizes and industries. More specifically, this data is based on a survey from the Small Business Administration conducted between 1990 and 1992. There are more recent surveys available but this one was selected because there was a recession during this time, which will ensure that the employer/employee dynamics are most similar to those that currently exist in the labor market. Accessed via: Barron J.M., Berger M.C. and Black D.A. “Employer Search, Training, and Vacancy Duration.” Economic Inquiry, 35 (1997): 167–192.