By Brian Points
EMSI Consulting Economist
For the past few years, wage growth across the U.S. has been minimal, even among occupations that are allegedly in high demand. This might seem odd given how often employers say that finding and retaining good talent is a key issue for their business operations. Does this mean firms are not responding to the skills gap? Not necessarily. Companies deal with increased demand for workers in different ways, which will likely take time to actually register in wage data. In addition, limitations of the United States’ occupational classification system also may shroud some trends that are actually occurring.
EMSI conducted new research into this issue and found that jobs related to health care, the internet and software, finance, and engineering have seen notable wage increases since 2006. However, the well-known fact still stands – there does appear to be an actual shortage of qualified candidates available for these positions.
Analysis of wage changes and the potential supply of workers indicates there are 1.9 available workers for every opening in computer occupations – the lowest ratio for occupation groupings with notable wage growth. The available workers-to-openings ratio is similarly low for health diagnosing and treating practitioners (2.3), financial specialists (3.1), health technologists and technicians (3.1), engineers (3.1), and occupational therapy and physical therapist assistants and aides (3.4).
Available Workers-To-Openings Gap Ratio of Occupations with Wage Growth
|3-digit SOC||US Gap Ratio|
|Health Diagnosing and Treating Practitioners||2.3|
|Health Technologists and Technicians||3.1|
|Occupational Therapy and Physical Therapist Assistants and Aides||3.4|
The same analysis was also performed for the state of Oregon because of the availability of some key information required for regional analysis. This analysis reveals skill gaps in the same occupation groups, but these issues may be slightly more pronounced among computer occupations, engineers, and some healthcare workers.
Data Used and Key Considerations
EMSI’s latest research on the skills gap uses four factors:
- Inflation-adjusted wage growth between 2006 and 2011,
- Completers of certificate and degree programs,
- Unemployment by occupation, and
- New and replacement jobs projected over the next five years.
These data sources help illuminate possible skill gaps, but data limitations and common misconceptions about how businesses operate are important to consider in any skills gap discussion.
In particular, two issues warrant special attention:
- Wage changes take time: A large proportion of employers claim that a lack of skilled workers affects the quality or amount of products or services that they are able to produce. So why wouldn’t a company just raise wages for workers that are most needed? Because it is not so simple to change wages of workers within a company. Most firms’ budgets are carefully crafted to ensure that funding represents the value that is returned to the company through each investment. Therefore, to raise wages in one department, funding must be cut somewhere else. This partly goes into what economists call “sticky wages” — that is, the condition where companies do not quickly move wages either up or down although market forces indicate that they should. Given this explanation, we should not expect to see wage changes occurring over a short period of time, say one year. Rather, if demand for certain types of workers truly is increasing, we should expect to see wages shifting over an extended period of time.
- Data sources may not adequately capture actual wage changes: Because of aggregation, mislabeling, and other issues, data is imperfect and imprecise – even for the most detailed occupations. When we analyze data for a particular occupation code, we are implicitly saying that average wages for an occupation are a fair approximation for the value of all workers in that SOC category. But for any one SOC code there is wide variation in education, experience, and skills, which also means that there is a wide variation in wages. Moreover, there are the different types of businesses that could possibly hire that worker, and a range of what they can afford to pay.
A full discussion of these topics and the findings from EMSI’s data analysis are available in the following white paper.
The skills gap refers to a structural misalignment between education, workforce development, and the labor market. Discussion of the skills gap has been particularly active since the Great Recession, as unemployment rates have remained stubbornly high despite increases in job openings. A skills gap can exist in any industry or occupation, but typically it is discussed in a handful of fields including skilled manufacturing workers, healthcare workers, and science, technology, engineering, and mathematics (STEM). The issue touches at the heart of both workforce and economic development. Workforce developers do not want to miss opportunities to guide people toward high-quality jobs, and economic developers do not want growth in their local economies to constrained by a lack of the right types of workers.
EMSI first contributed to the skills gap conversation in 2011 with two blog posts (“Dealing with Skills Shortages in a Complex Economy” and “A Detailed Look at Skills Shortages”). Since that time, the discussion has continued to gather steam and yet no consensus has been reached regarding the cause of this complex problem. Indeed, commentators seem to have fractured into several different camps that each has their own explanations for the problem. In this post we will explore some of the common explanations and offer a new look at some data at both the national and the regional level.
Thoughts on Some Common Explanations
In recent years, countless sources have reported businesses in the United States having trouble finding workers with the right skills sets. For example, the 2012 Manpower Talent Shortage Survey asked respondents whether they had difficulty filling jobs. Among industrialized countries the United States ranked fifth highest with 49% of companies responding yes to this question. Furthermore, when respondents were asked to report their most common reasons for difficulty filling job openings 55% of American employers name “lack of available talent/no applicants” and 44% name “lack of technical competencies (hard skills).” On the other hand, the national unemployment rate has yet to drop below 7.5% since the end of the recession more than three years ago. With 11.8 million unemployed people in the United States, clearly the issue is not a lack of people who need work. One possible explanation is that people who need work lack the requisite skills to obtain a job. Another explanation, which has been propagated more recently, is that employers are not offering enough money to entice potential employees, who are rather choosing to continue in their current positions or remain unemployed until a better offer comes along.
The thinking makes sense — if you can’t find enough people, or the right people at a given wage, you must raise that wage to attract more and better workers. Though this explanation makes perfect microeconomic sense, either employers are not practicing it or we can find little evidence of it happening in the available data. Both explanations are true to some extent.
Wage Changes Take Time
Skills shortages are no trivial issue. A large proportion of employers claim that a lack of skilled workers affects the quality products or services, or the amount that they are able to produce. For instance, the Manpower survey reveals that 44% of employers believe that unfilled positions will impact their customers or investors. Another survey shows that 47% of manufacturing businesses in the U.S. worry about maintaining production levels to meet customer demands. So why wouldn’t a company just raise wages for workers who are most needed? It seems simple enough from an outside perspective that employers could simply reallocate resources toward parts of their companies that are most productive, but it turns out, it is not so simple to change wages of workers within a company.
The misperception might partly stem from a misunderstanding about how firms operate. It might be tempting to think that companies, especially big corporations, are sitting on piles of money that they can easily dole out if they only cared enough. In reality, most businesses function more efficiently than that. Their budgets are carefully crafted to ensure that funding represents the value that is returned to the company through each investment. Therefore, to raise wages in one department, funding must be cut somewhere else.
We should not expect to see wage changes occurring over a short period of time, say one year, but if demand for certain types of workers truly is increasing, we should expect to see wages shifting over an extended period of time.
Consider this: If you were the CEO of major manufacturing company, would you want to reduce commissions for salespeople? Decrease the marketing budget? Maybe, but if you make any drastic changes, you should expect some negative consequences. By lowering wages for an entire group of workers, morale would also be lowered, which could lower productivity. You could also expect some attrition, especially if you lower wages below market averages, as workers would jump ship for higher-paying positions. Also, because many business commitments are for year or multi-year periods, you cannot shift money around as quickly as you would like (e.g., you’re stuck in a building lease, have a multi-year contract with a supplier, etc.) As the CEO, you know that not having skilled tradespeople on your shop floor will eventually decrease your bottom line, if it hasn’t already. One option is to increase wages over time by allowing higher annual raises, percentage-wise, than what is being given to people in other departments. This will allow your company’s budget to slowly equalize with what workers are contributing to the company. Another option is to lay off some workers, so you can afford to raise wages for those workers who are most needed. Both options are exercised by companies from time to time, and each has their advantages and disadvantages.
The complications involved in wage changes are endless and we need not cover all the details here. This outline is enough to show that reallocating labor costs within a single firm is a rather complicated issue. This forms part of the explanation for what economists call “sticky wages” — that is, the condition where companies do not quickly move wages either up or down although market forces indicate that they should. Given this explanation, we should not expect to see wage changes occurring over a short period of time, say, one year, but if demand for certain types of workers truly is increasing, we should expect to see wages shifting over an extended period of time. This leads to the next question: are wages increasing over time for occupations in high demand?
Data Sources May not Adequately Capture Actual Wage Changes
Those of us who deal with data on a regular basis can sometimes fool ourselves into thinking that we can find an answer to virtually any question that is posed to us. In this case, the data can only take us so far. The federal government collects employment and earnings information for over 700 standard occupational classification (SOC) codes, but even at that precise level a great deal of detail is being swallowed up by aggregation.
Those of us who deal with data on a regular basis can sometimes fool ourselves into thinking that we can find an answer to virtually any question that is posed to us. In this case, the data can only take us so far.
When we analyze data for a particular SOC code we are implicitly saying that average wages for an occupation are a fair approximation for the value of all workers in that SOC category. But for any one SOC code there is wide variation in education, experience, and skills, which also means that there is a wide variation in wages. Moreover, there are the different types of businesses that could possibly hire that worker, and a range of what they can afford to pay. The labor market is not like a shoe store, where each shoe type is stitched and printed to be a facsimile of all other shoes of that type and brand. Each individual that a business could hire is different in terms of skills, experience, and interests. In summary, being stamped with a SOC code by one human resources department does not mean that a worker has the same skill set, and therefore deserves the same wage, as any other person stamped with that SOC code.
This issue can be especially problematic in occupations and industries undergoing major technological changes. Take, for example, mechatronics technicians, whose skills in programming, repair, and production are in high demand with many advanced manufacturing businesses. As manufacturing becomes more robotic and less manual, such workers are critical for programming and operating machines to do work that used to be done by humans. Since there is no SOC code for “mechatronics technicians,” they are categorized in countless related SOC categories, making it very difficult to collect accurate data on these workers, and muddying the data for all related SOC categories. The most likely categorization for mechatronics technicians is electro-mechanical technicians (SOC 17-3024), an occupation that has decreased in employment and only marginally increased in wages over the past few years.
Figure 1 uses hypothetical data to display how wage growth for mechatronics technicians as a sub-category of electro-mechanical technicians can be obscured by aggregation. Though the wages numbers are only hypothetical, they are based on observations of wage differentials for workers with mechatronics skills in Pennsylvania. In this example, mechatronics technicians start the time period with higher wages than other electro-mechanical technicians and are seeing above average wage growth whereas other electro-mechanical technicians are seeing lower wage growth. Yet, since the majority of workers classified as electro-mechanical technicians are not mechatronics technicians, wage growth for the SOC code 17-3024 appears rather underwhelming.
Since demand for mechatronics technicians is high, in the long-term we should expect that subcategory to grow, allowing the aggregate wage data to better reflect businesses high demand for mechatronics technicians. In the meantime, however, we may not see the growth that we expect in the aggregate data.
An Update Using National and Regional Data
The long preface to the data analysis portion of this post allows us to approach the data with the proper context. Given what has been outlined, we expect there to be some wage growth, but it will be gradual. And there may be many fields where labor market demand is increasing, but it is not borne out in wage data because of aggregation issues.
For this data analysis, we looked at real wage growth for 3-digit SOC codes in the nation and determined which ones indicate signs of increased demand. After highlighting those occupational groups, we looked at the supply of potential workers at the national level and for the state of Oregon to determine whether these occupations are being fed with the necessary number of workers to meet labor market demand.
In our original series of posts, we looked at occupations at the 5-digit SOC level that have seen considerable growth in both employment and wages over an 11-year period. The analysis was based on the expectation that occupations with sustained high demand should see above-average growth in both of these categories. The second analysis looked at the supply side of the equation, analyzing potential employees through recent educational program completers, the unemployed, and people with compatible skills. A ratio of “potential workers” to annual openings was provided for each occupation, which was compared to the national average.
This clustering pattern, with few occupations lagging far behind or skyrocketing ahead of the others, provides justification for the two claims made earlier: earnings by occupation will change slowly, and much of the actual wage change going on will be difficult to detect.
For this update, several aspects of the analysis were dropped or modified. First, doing the analysis at the regional level forced us to look at occupational data at the (more general) 3-digit SOC level rather than at the (more specific) 5-digit SOC level. Secondly, we only analyzed wage growth to determine potential skills shortages, rather than wage and employment growth. One thing that the skills gap discussion has revealed is that for many occupations, supply simply cannot keep up with demand; therefore we cannot expect employment to continue rising even if employers are demanding more and more workers. Lastly, the compatible workers numbers were not reproduced, as it was very difficult to develop accurate wage criteria for determining why workers choose to abandon one occupational category and enter another.
In summary, this updated analysis takes in account the following data elements for the United States and Oregon:
- Inflation-adjusted wage growth between 2006 and 2011 (just for the United States),
- Completers of certificate and degree programs,
- Unemployment by occupation, and 
- New and replacement jobs projected over the next five years
Demand Side Findings – Wage Growth
We must first look at wage trends in the national economy to see how many occupational categories have displayed a growth rate discernibly higher than average. We analyzed earnings for 94 3-digit SOC categories and ranked the occupational categories according to average annual growth between 2006 and 2011. Analysis of wages was done at the national level only and not for the state of Oregon because availability of wage data from year-to-year at the state level is limited. Furthermore, wages trends in such large geographic areas tend to follow national trends.
As Figure 1 indicates, most 3-digit occupations saw moderate annual growth, with only a handful measuring exceptionally high or exceptionally low. Average change was near zero at 0.1% and the preponderance fell within 0.7 percentage points of the mean (-0.6% to 0.8%). This clustering pattern, with few occupations lagging far behind or skyrocketing ahead of the others, provides justification for the two claims made earlier: earnings by occupation will change slowly, and much of the actual wage change going on will be difficult to detect. That being said, the occupations with above average growth are worth investigating as they may indicate some increase in labor market demand, relative to other occupational groups.
To filter out extremely low-skilled positions, we focused just on those occupational groups with average wages between 2006 and 2011 that are greater than $26,000. Within this group, there are 41 occupational groups that saw average annual growth greater than 0.1%. A list of the more notable occupations in the group are shown in Table 1.
Table 1: Notable Occupational Groups with Above Average Wage Growth
|3-digit SOC||Description||Average Annual Wage Growth|
|31-2||Occupational Therapy and Physical Therapist Assistants and Aides||1.50%|
|29-1||Health Diagnosing and Treating Practitioners||0.50%|
|17-1||Architects, Surveyors, and Cartographers||0.50%|
|29-2||Health Technologists and Technicians||0.30%|
|17-3||Drafters, Engineering Technicians, and Mapping Technicians||0.20%|
|31-9||Other Healthcare Support Occupations||0.10%|
A number of occupations that have remained hot topics in the skills gap discussion are contained in this list, including four from the healthcare sector: occupational therapy and physical therapists assistants and aides (31-2), and health diagnosing and treating practitioners (29-1), health technologists and technicians (29-2), and other healthcare support occupations (31-9). Also included are a host of STEM occupations: physical scientists (19-2), life scientists (19-1), architects, surveyors, and cartographers (17-1), engineers (17-2), computer occupations (15-1), drafters, engineering technicians, and mapping technicians (17-3) and financial specialists (13-2).
Perhaps the most often-cited occupational category in the skills gap discussion is skilled manufacturing workers. There were also a number of manufacturing occupations with above-average growth, and just as many that did not surpass the 0.1% wage growth threshold. Those in manufacturing that did appreciate in wages include workers in management positions or in metal and plastics manufacturing. Alternatively, those that did not increase or decreased in wages are occupations in older-line forms of manufacturing where many jobs are being replaced with labor-saving equipment, such as woodworkers and food processing workers.
Table 2: Wage Change Among Manufacturing Workers
3-digit SOC Description Average Annual Wage Growth
Above Average Growth
51-8 Plant and System Operators 0.70%
51-1 Supervisors of Production Workers 0.30%
51-4 Metal Workers and Plastic Workers 0.20%
At or Below Average Growth
51-7 Woodworkers 0.00%
51-2 Assemblers and Fabricators -0.10%
51-6 Textile, Apparel, and Furnishings Workers -0.20%
51-3 Food Processing Workers -0.50%
Supply Side Findings – Available Workers
A notable pattern on the low end of the list is the appearance of occupations in service categories. For example, four of the largest service-producing categories in terms of 2011 employment all experienced below-average wage growth, which includes retail sales workers (41-2), other sales and related workers (41-9), and information and records clerks (43-4).
Examining wage change data can point the compass toward occupations that employers seem to be favoring relative to others in the labor market, but it does not prove that a skills gap exists. Economic theory suggests that as wages rise in a particular occupation, more and more people will be attracted to that line of work, thereby satisfying the gap that previously existed. For this reason we also must examine sources of supply to determine whether occupations experiencing wage growth are being sufficiently provided with qualified workers.
For this side of the question EMSI looked at two of the most likely sources of supply – unemployed workers and educational program completers – to determine the stock of available workers for each 3-digit SOC code. The sum of these two sources is compared to our forecasted average annual openings (an estimate of new and replacement jobs). We used the ratio of available workers to openings, a number we are calling the “gap ratio,” to determine supply in that occupational category. Given what we can easily find out about these occupational categories this is a reasonable metric for determining the strength of supply.
Gap Ratios for Occupations
National Gap Ratios
Using the gap ratio we are able to determine an average gap for 3-digit SOC codes. What we cannot predict is the “right” ratio of supply and demand. That question depends on issues such as the quality of each person’s education, their values in looking for a job, and many other factors. Consequently, these data cannot tell us which occupations are under and oversupplied, but what we can conclude is that occupations with larger gap ratios are more abundantly supplied, and occupations with smaller gap ratios are less abundantly supplied.
We will focus on occupations displayed in Table 1, which are those with notable wage growth outside of the manufacturing sector. Since one of the key sources of supply is educational completers, the method does not work as well for occupations that do typically require a college degree or certificate, such as the manufacturing occupations shown in Table 2.
By merging the supply and demand data at the national level, we can see that of the 11 highlighted occupations, five have gap ratios lower than the national median: engineers, health technologists and technicians, health diagnosing and treating practitioners, finance specialists, health diagnosing and treating practitioners, and computer occupations. Before explaining the implication of these data it is necessary to explain which specific occupations belong in each of these categories. The two healthcare categories encompass over 70% of all healthcare workers, including doctors, registered nurses, and other types of diagnostic technicians. The computer occupations category is also fairly comprehensive, including IT types such as network administrators and development types such as software engineers. The finance specialists category is dominated by accountants and auditors, in addition to several other high-level finance workers. Engineers includes all forms of engineers, from mechanical to nuclear. What these figures reveal about the American labor market is that, on the whole, healthcare occupations, computer occupations, financial specialists, and engineering occupations have seen notable increases in wages, yet there are still not as many workers available for these positions as for most others in the labor market.
Most of the other six occupations above the median do not seem to be considerably oversupplied, spanning from a low of 3.4 to a high of 13.7. By contrast, one of the highest-ranking occupations on the list, social scientists and related workers, has a gap ratio of 39.6. The group of other healthcare support occupations is an interesting contrast from the other healthcare occupations with lower gap ratios. This group contains many relatively low-skilled healthcare workers such as dental assistants and medical assistants that typically require short-term education and/or a postsecondary certificate. Programs training for these occupations have ballooned at community colleges in recent years, as many people were attracted to wages in the healthcare industry but were not prepared to spend several years receiving an education. The high gap ratio for healthcare positions requiring less than an associate’s degree could soon send a signal to employers that wages have reached a point at which more than enough workers are available.
The two science categories on this list – physical scientists and life scientists – may not be a perfect fit with this model of analysis. Most jobs in science fields require an advanced degree, but on the supply side we have counted all those receiving bachelor’s degrees as potential workers. Notwithstanding the methodological limitations, this could still point to an educational misalignment issue. For example, in 2011 more than 139,000 people in the United States completed some form of education program in life sciences (69,000 of which received a bachelor’s in biology), but there are only about 11,000 openings per year that require a degree in life sciences. The fact that wages are still rising for these occupations is likely driven by the fact that the much smaller pool of people with advanced degrees in life sciences are not abundant enough to maintain wages at a stable level.
Table 3: Gap Ratio of Occupations with Notable Wage Growth in US
|3-digit SOC||US Gap Ratio|
|Health Diagnosing and Treating Practitioners||2.3|
|Health Technologists and Technicians||3.1|
|Occupational Therapy and Physical Therapist Assistants and Aides||3.4|
|Architects, Surveyors, and Cartographers||5|
|Drafters, Engineering Technicians, and Mapping Technicians||6.2|
|Other Healthcare Support Occupations||7.4|
Oregon Gap Ratios
Although the median gap ratio for Oregon is much lower than the national benchmark (2.1 compared to 3.4), we elected to use the national benchmark for comparison because it is likely that the United States better approximates a closed system where supply and demand should be roughly equal. The 2.1 figure could be too low to use as a target, as it seems Oregon imports more workers from outside of its borders than does the United States.
Examining the data for Oregon reveals patterns that are quite similar to the nation, but the gap ratio is lower among most occupational groups. The same five occupations have gap ratios below the national median, with the addition of occupational therapy and physical therapists assistants and aides. These data indicate that the state of Oregon experiences many of the same skills gap issues as the nation as a whole. However, the problem is likely a bit more severe among computer occupations, engineers, and some healthcare workers.
Table 4: Gap Ratio of Occupations with Notable Wage Growth in Oregon
|3-digit SOC||Oregon Gap Ratio|
|Health Technologists and Technicians||2.3|
|Health Diagnosing and Treating Practitioners||2.4|
|Occupational Therapy and Physical Therapist Assistants and Aides||3.3|
|Drafters, Engineering Technicians, and Mapping Technicians||4.3|
|Architects, Surveyors, and Cartographers||6.6|
|Other Healthcare Support Occupations||6.7|
Wage growth by occupational sector has been minimal over the last few years, but this is not necessarily a sign that firms are not responding to the skills gap. Firms deal with increased demand for workers in many different ways, which may take some time to register in commonly used wage data. Furthermore, the limitations of our occupational classification system may shroud some trends that actually are occurring. Despite these restrictions, careful analysis of wages changes and the potential supply for workers indicates some areas where the skill gap could be particularly acute. These include healthcare occupations, computer occupations, financial specialists, and engineering occupations. Analysis of the state of Oregon reveals skill gaps in the same occupational groups, but these issues may be slightly more pronounced among computer occupations, engineers, and some healthcare workers.
 Manpower Group, “2012 Talent Shortage Survey: Research Results.” http://www.manpowergroup.us/campaigns/talent-shortage-2012/. Accessed March 1, 2013.
 A third explanation that is not covered in this post is that there really are not any skill gaps but that the extension of long-term unemployment benefits has allowed workers to stay unemployed for longer with little fiscal consequences. This theory is outlined well in a paper by Rand Ghayad and William Dickens titled “What Can We Learn by Disaggregating the Unemployment-Vacancy Relationship?” Their outline provides some possible explanations for why unemployment has remained high, but does not adequately dismiss the Skills Gap possibility. People choosing to remain unemployed may be basing their choice at least partly on the fact that they do not have the skills that they need to find a job that they want.
 Deloitte and The Manufacturing Institute “Boiling Point?: The skills gap in U.S. Manufacturing”
 Morale is the most likely explanation for slow wage change according to one economist. Truman Bewley, Why Wages Don’t Fall During a Recession.
 Thanks to Scott Sheely at the Lancaster County Workforce Investment Board in Lancaster, Pa., and the Pennsylvania Center for Advanced Manufacturing Careers.
 All wage numbers cited in this report have been adjusted for inflation and converted to 2011 dollars using the CPI index.
 There reasons for this are twofold. One, it is not possible to perform a regional analysis at the 5-digit level because precise unemployment information is not available for sub-national areas. And two, the BLS updated the SOC coding system in 2010 which made substantial alterations to 5-digit SOC categories. Analysis of wage changes relies on consistent data categories over time so analysis at the 5-digit level would be very specious.
 It is still valid to look at both wage and employment growth but for a more specific subset of occupations – that is, those that have not yet encountered any supply availability issues.
 Although Oregon only publishes unemployment at the 2-digit level. EMSI used national distribution of unemployment from 2- to 3-digit SOC to estimate the number of unemployed in the state of Oregon. These numbers were further regionalized by accounting for the statistical relationship between unemployment and job change between 2010 and 2011.
 This is driven by workers’ desire to find the best wage for their position. If wages for an occupation were growing at an unusual rate in a particular area, people would flood to that area from other states, which would cause employers to bring the wage back down closer to the average national rate.
 Technically speaking, 68% of all observations fall within one standard deviation of the mean.
 We do not call this a supply-demand ratio because that would imply that all potential sources of supply are being considered. We recognize that there are a number of sources not being considered such as people with compatible skills who could be enticed to switch occupations, immigrants, and currently retired workers who could reenter the workforce.
Emsi turns labor market data into useful information that helps organizations understand the connection between economies, people, and work. Using sound economic principles and good data, we build user-friendly services that help educational institutions, workforce planners, and regional developers (such as WIBs, EDOs, chambers, utilities) build a better workforce and improve the economic conditions in their regions.