NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

National Research Council (US) Center for Education. Research on Future Skill Demands: A Workshop Summary. Washington (DC): National Academies Press (US); 2008.

Cover of Research on Future Skill Demands

Research on Future Skill Demands: A Workshop Summary.

Show details

6Skill Supply and Demand

On the second day of the workshop, the focus shifted from skill demand to the national supply of skill, and the balance between supply and demand. Arne Kalleberg opened the session by reflecting on the previous day’s discussion of skill demand.


If current trends continue, Kalleberg said, job growth will be concentrated in high-wage professional, technical, and managerial jobs and in low-wage service jobs, raising questions about future prospects for upward career mobility. He called for further study of possible growth in mid-level jobs, noting that Genentech Corporation is an example of increased employment of mid-level technicians (Murray and Hsi, 2007a).

Kalleberg argued, however, that these current trends may not continue, as future changes in jobs will be affected by social choices and employer decisions about how to organize work. He observed that the diversity of service work—including highly paid knowledge workers, low-wage service workers, and technoservice workers who combine both types of work (Darr, 2007b)—highlighted a key point: service work can be performed at different levels of quality (see Chapter 4). “If we decide as a nation to take better care of our citizens, our elderly, our children, and are willing to pay for it,” he said, many low-level service occupations might be transformed to mid-level occupations.

Kalleberg said that a key question from the previous day is the extent to which future work will be performed in the United States or offshore. He noted that Autor had suggested that routine jobs that could be digitized were most likely to be outsourced (Autor, 2007a), but added that regulatory mechanisms—which reflect political and social choices—would also be likely to affect the extent of outsourcing (Levy and Goelman, 2005).

Commenting on disagreements about the extent to which different categories of jobs—such as “routine” jobs or “service occupations,” required higher or lower skills, Kalleberg said that these reflected deeper disagreements about how to define “skill.” He observed that the Bureau of Labor Statistics (BLS) examines skill in terms of wage levels, education and training requirements, and the educational attainment of the labor force (Sommers, 2007). He described the O*NET database as a “positive step” toward improved definitions of skill, and observed that Handel had developed new direct measures of skill for his survey (see Chapter 5). Asserting that definitions of skill remain “contested” (e.g., Attewell, 1990), Kalleberg argued that, without agreement on an overall measure of skill, it might not make sense to ask whether the skill demands of work are increasing or decreasing; he suggested asking instead about how different dimensions of skill may be changing.

Kalleberg observed that one of the most important themes of the previous day is that society constructs workplace skill demands, through choices about how to package tasks together into jobs and occupations. He called for greater awareness of this theme, recognizing that not only labor markets, but also workers and skill definitions, are adaptable. He then reflected on debates about the factors determining wages for various jobs, including skills, the supply of labor, institutions, and other factors, noting that the research is unclear on this point.

In contrast to these many areas of uncertainty and disagreement, Kalleberg said he had observed much agreement about another important topic—the kinds of competencies and skills that would be required in the future. He repeated the list of skills that Houston presented, including creative problem solving, complex communication skills, adaptability, self-management, personal development, and systems thinking. Kalleberg asserted that these skills would be likely to be required in all kinds of jobs, especially if policies are created to upgrade the skills of low-wage jobs. He suggested—rather than researching the links between this list of skills and changes in the workplace (he noted the example that self-management is important because many firms now employ fewer supervisors)—taking the list “as a given.” Kalleberg suggested that education and training policies focus on this list of broad competencies as the kinds of skills needed in the future labor force, while also continuing to teach basic skills of reading, mathematics, and thinking.


Kalleberg then moderated a panel discussion of skill supply and demand, including labor economists Peter Cappelli and Harry Holzer and sociologist B. Lindsay Lowell. Each panelist was invited to make brief opening remarks, followed by discussion among the panelists and then discussion with the audience.

Opening Remarks

Peter Cappelli: Is the Future Skill Supply a Public Policy Problem?

Cappelli argued against using input-output models to approach skill questions as if they were engineering problems, saying that such an approach might suggest that “inexorable” developments in technology would require “a particular level of skill.” He said that the earlier workshop discussions had shown that social and economic choices had a strong influence on skill requirements. As further evidence of the influence of social and economic choices, he said that researchers had found that numerically controlled machine tools affected skill demands quite differently in different settings, depending on management choices, the power of unions, wage levels, and other factors (Shaiken, 1986; Adler and Borys, 1989).

Cappelli noted that the earlier discussions had not focused on the supply and price of labor at different skill levels, and the relationship between these variables. He said that a large supply of low wage labor influences employers’ choices about how to organize work, including its skill demands. To illustrate this point, he described his experience working in Bahrain, United Arab Emirates. He observed white-collar professionals working in modern office buildings using laptops, just as professionals do elsewhere in the world, but also saw construction workers carrying cement on their heads and digging holes with shovels, lacking any modern technology. Cappelli said that the contrast can be explained by the nation’s guest worker program, which provides very low-skilled labor from South Asia. As a result, the price of construction labor is very low, and construction firms do not need to invest in cement mixers, tractors, or other technology. Summing up his point, Cappelli said that wage levels affect employer’s choices about skill requirements, and the wage levels are “very much a function of labor supply.”

Next Cappelli addressed the common argument that the U.S. labor force is shrinking. He presented population profiles from several countries (U.S. Census Bureau, 2007), noting that the population of most industrialized countries in Southeast Asia is expected to shrink dramatically by the year 2025 because of declining birth rates. Presenting a population profile for Spain, he observed that the populations of most European nations are also expected to shrink rapidly because of large declines in birth rates. In contrast, the population of Afghanistan and many other Arab countries is expected to grow rapidly. He then presented the U.S. population pyramid, stating that it is more evenly balanced than either of those two groups of nations and that the United States is the only developed nation with a growing population. Cautioning that population is not the same as labor supply, he said that BLS projects that the growth rate of the labor force will begin to slow down. He said that the most recent projected growth rate (an average of 1 percent over the decade 2004-2014) is only 0.2 percent lower than the growth rate from 1994 to 2004 (Toossi, 2004), describing this as a very small change, “particularly given how imprecise these estimates are.” Cappelli went on to argue that some workers are retiring later than in previous generations, at age 68, and that it is “very easy” for people to return to work after retiring.

Cappelli said he had become interested in the question of whether the national labor supply is adequate because so many employers told him they were concerned about shortages (Cappelli, 2003). He refuted the idea that a nation requires “a bigger labor force in order to grow,” saying that the U.S. economy is now 8 times larger than during World War II, but with a labor force only twice as large. He noted that simple input-output models that fail to fully account for productivity growth—such as those used by BLS—would require the U.S. labor force to be four times as large as it was to sustain the nation’s level of economic activity at the time of the workshop.

Going on to what he described as his “punch line,” Cappelli said that employer concerns about possible labor shortages are not completely erroneous but reflect the fact that employers are no longer developing skills internally (Cappelli, 2007). He noted that Osterman had done some of the early research on internal labor markets (Osterman, 1984), showing that companies that had formerly hired entry-level people and trained and promoted them over time no longer did so. Cappelli cited a study by Taleo, a manufacturer of employment software, which found that, among firms using its software, about two-thirds of all job vacancies were filled by outside hires (Cappelli, 2007). Cappelli said that the trend of outside hiring, rather than internal development and promotion, is most apparent in Silicon Valley, where technology firms rely almost entirely on colleges and universities to provide the specific skills they need (National Research Council, 2001).

Cappelli concluded his description of the current balance of skill supply and demand by asking whether or not it is a policy problem. He asserted that “employers would clearly like it to be a public policy problem,” making the education system responsible for developing the skilled employees they require. He recalled that employers expressed similar concerns in the 1980s, despite higher unemployment rates at that time, because they were no longer prepared to train new employees. He argued that employers’ current concerns reflect their demand for employees with the right attitudes, discipline, and “the skills to step immediately into the job and start doing the work.” Describing this demand as a “huge and quantum paradigm change,” Cappelli asked whether policy makers should assume responsibility for meeting it.

Harry Holzer: Possible Imbalance in Skill Supply and Demand

Holzer said that, although he took concerns about the future skill supply more seriously than Cappelli, he agreed that attempts to accurately project demand for skills and supply of skills in order to compare them and map areas of shortage are “wrong, a completely noneconomic way of thinking.”

Holzer said that two trends—the retirement of baby boomers and immigration patterns—could cause problems in the future national skill supply (Holzer and Nightingale, 2007). He predicted that, over the next 20 to 30 years, the retiring baby boomers would be replaced by immigrants who, on average, had “considerably less formal education and training and skills.” Taken together, these two trends imply that the “aggregate quantity of skills in the labor market is declining,” he said. Holzer cautioned that his prediction assumed that current trends in educational investment, immigration, and retirement would all remain unchanged. He challenged Cappelli’s arguments about increased working among older workers, noting that adjustments in retirement behavior varied by income level and educational attainment (Abraham and Houseman, 2007). Holzer said it is easier for educated professionals to continue working into their seventies than for those who had worked in construction or other physically demanding jobs.

Holzer said he expected that, because of these two trends, there would be some slowdown in the supply of skills and that, if demand continued to grow, there would be “some imbalance in the labor market.” Acknowledging that the labor market would adjust to any imbalance, he argued that the adjustment did not mean that “there is no issue and no problem.” Holzer went on to suggest that the growth of income inequality over the previous 30 years is due not only to institutional and organizational factors, but also to growing demand for skills, so that a possible future slowdown in the skill supply might lead to still further increases in inequality. He expressed concern that the rising costs of higher education are making it more difficult for students from low- and middle-class families to graduate, predicting that continued increases in college costs would exacerbate inequality.

Turning to skill demand, Holzer observed that the earlier discussions had highlighted uncertainties about the future. To illustrate the uncertainties, he noted that Richard Freeman had recently written a paper arguing that any future declines in skill supply due to baby boomer retirements would be more than offset by movement of jobs (and skill demand) offshore (Freeman, 2007). Holzer questioned Freeman’s conclusion, referring to an analysis by economist Edward Leamer (2007), which refutes Thomas Friedman’s argument that the world is flat (2005). Holzer said that, while there is likely to be an increase in movement of work offshore, arguments about a “massive flight of skilled jobs” are probably exaggerated.

Holzer went on to say that, despite these uncertainties, he expects a “modest slowdown” of skill supplies, and he ventured several guesses about the future. First, he predicted that there would be “major pockets of tightness in certain labor markets,” especially when jobs were difficult to outsource and could not easily be filled by immigrants. He said this included some health care, skilled construction, and manufacturing jobs, observing that these are mid-level jobs that would generate large replacement demand as the baby boomers retired over the coming decades. Holzer said that immigrants would be unlikely to fill these mid-level jobs, because most immigrants are either highly skilled or low skilled. He suggested there might be long-term shortages of skilled workers in a range of health care occupations, because (as Peter Kemper had mentioned) federal reimbursements constrain costs and wages in this sector.

Second, Holzer predicted that inequality in the labor market would continue and possibly increase. He reiterated his view that increases in inequality are due in part to a lack of skills at the low end of the labor market, such as among high school dropouts. He said that the lack of broad competencies—not only among dropouts but also among some high school graduates—is clearly constraining these individuals’ ability to advance in the labor market. Holzer cautioned that data on the labor market experiences of disadvantaged young men are weak because the Current Population Survey and the decennial census do not reach all of them (Holzer, Edelman, and Offner, 2006). He also repeated his earlier concern that the rising costs of higher education might exacerbate inequality in the labor market.

Third, Holzer said that employment insecurity may grow in the future. While cautioning that the data are mixed, making it difficult to “make a strong claim,” he predicted that outsourcing of jobs and creation of new jobs with different skill demands could reduce employment security. He suggested three types of policies to address different aspects of insecurity. First, to help people adjust to possible future changes in jobs and skill demands, he mentioned policies to help young people and working adults develop the broad competencies discussed earlier. For example, policies could support community college programs for upgrading the literacy, broad competencies, and technical skills of adults. Second, predicting that there would continue to be many low-wage workers, he suggested a mix of policies to increase their earnings, as others had recommended (e.g., Osterman, Kochan, Locke, and Piore, 2001). For example, low-wage workers’ incomes could be supplemented through the earned income tax credit. He suggested combining “sticks,” requiring higher wages—such as a higher minimum wage, increasing unionization, and collective bargaining—with “carrots,” such as tax credits and technical assistance. This combination of policies, he said, might induce more employers to train their workers and pay them higher wages.

Holzer concluded by advocating a third type of policy to help workers in the face of increased insecurity—federal insurance systems. He suggested “lifelong learning” insurance to help adults who need retraining due to a job change, improved unemployment insurance, and universal health insurance not tied to employment. Finally, he suggested a system of federal wage insurance to help reduce the loss of wages that workers may experience when changing jobs.

B. Lindsay Lowell: Future Immigration Trends

Lowell (Georgetown University) opened with an overview of the current immigration system, including permanent visas, temporary visas, and illegal immigration (Lowell, 2007). He observed that about 800,000 people receive permanent visas each year, with a backlog of about 2 million people waiting for such visas. Another 500,000 to 800,000 people enter each year on temporary visas, including H1-B visas for skilled workers and H2-B visas for less skilled agricultural workers. Between 500,000 and 700,000 illegal immigrants also enter the United States each year. Noting that most of the illegal immigrants lack a high school education, whereas most temporary visas and some permanent visas are given to highly educated workers, Lowell agreed with Holzer that the educational attainment of the immigrant labor force is a “barbell distribution, a little heavier on the bottom than the top.”

Lowell said that many highly educated immigrants are employed in science, technology, engineering, and mathematics (STEM) careers, in contrast to the popular perception of immigrants as unskilled. Between 1990 and 2000, the immigrant share of all doctorates employed in science and engineering fields grew from 24 to 38 percent, he said, while the share of younger doctorates in these fields (under the age of 45) increased from 26 to 52 percent. Given these high percentages of immigrants in STEM fields, Lowell questioned employers’ claims that they need a larger supply of foreign-born STEM workers (e.g., Business Roundtable, 2005).

Turning to the question of future immigration levels, Lowell presented a chart depicting the growth of college-educated citizens in many different nations, noting the rapid growth in India, China, and Mexico (Lowell, 2007). He said that highly educated people with graduate degrees in STEM fields are very mobile, and that about 30 to 50 percent of all such STEM professionals worldwide are employed in the United States. Lowell suggested that immigrants would fill a large share of the vacancies in BLS projections of large and fast-growing occupations to the year 2014. For example, he found that, in the 15 fastest-growing occupations (as projected by BLS—see Hecker, 2005), immigrants would make up 45 percent of medical scientists, 27 percent of computer software engineers, 18 percent of personal and home care aides, 17 percent of database administrators, and 17 percent of postsecondary teachers.

In order to assess how policy changes might affect future levels of immigration, Lowell presented an analysis of a Senate bill debated in spring 2006 (Senate Bill 2611), which would have increased the number of visas available to highly skilled computing and engineering workers by five- to six-fold. Comparing his projections of the number of new immigrants who would use these visas with BLS projections, Lowell found that, if this number of visas became available, by the year 2017, there would be 20 percent more foreign-born computing and engineering workers in the United States than could be absorbed by the available jobs (Lowell, 2006). He said that Congress continues to debate bills that would increase immigration levels without any consideration of “best guesses about future demand.” For example, he noted that Congress is considering increases in visas for scientists and engineers, although BLS had decreased its 10-year projection of growth in STEM occupations from 50 percent in 2002-2012 (Hecker, 2004) to 30 percent in 2004-2014 (Hecker, 2005) and despite the fact that there has been no growth in these occupations over the period from 2000 to 2005.

Lowell then shared some projections by Jeffrey Passel of the Pew Hispanic Center (2003). He noted that earlier Census Bureau projections had assumed that immigration would drop slightly around 2010 and then resume growing from around 2020 through 2050, in response to rising demand associated with baby boomer retirements. He said that, after being surprised by the 2000 census and other data showing that immigration had increased much more than had been anticipated, most experts were now predicting higher levels of immigration. Lowell said that Passel’s baseline projections indicate that immigrants would drive about 85 percent of future population growth to the year 2050 and make up about 19 percent of the national population in 2050. He reminded the audience that this was an average across all occupations and industries and that the share of immigrants in some sectors (such as STEM) is likely to be much higher.

Focusing on the skill composition of the future immigrant labor force, Passel projected that the proportion with less than a high school diploma would remain stable, while the share with a college degree would grow dramatically, Lowell said. As a result, he said, the foreign-born share of all college graduates in the United States would show a “remarkable” increase, from 15 percent today to 20 percent in the year 2050.

In conclusion, Lowell addressed the important question of how employers and individuals would respond to possible future laws that would increase visa levels. He argued that the response would be positive and that immigration levels would increase, for several reasons. First, he said, employers always take advantage of any increases to hire relatively low-cost labor. Second, he said that research shows that immigrants are driven by family networks, a lower “reservation wage,” and willingness to accept working and living conditions that native-born U.S. workers might not. Because of these factors, Lowell said, he and other experts are projecting large increases in immigration, regardless of future demand. Third, he said that econometric models of immigration have found that wage gaps are a major driving force. He noted that wage gaps between the most developed nations (such as the United States) and the less developed nations have been growing and are likely to remain high, despite recent wage increases in China and India. Fourth, he argued that the “population gap” between the most developed and least developed nations would grow quickly. Reminding the audience of the population pyramids Cappelli had presented earlier, Lowell said that, as young people in less developed countries reached the “migrant prone” ages of between 20 and 30, they would come to the United States seeking work.

Finally, Lowell said that policy debates were focusing on how to liberalize and admit more immigrants without really addressing how to enforce immigration laws. Although policies that would better match immigrants to the available jobs might be desirable, he has observed very little consideration of such policies.

Panel Discussion

Cappelli reiterated his contention that the retirement of the baby boom generation would not cause the U.S. economy to collapse, noting that the high school class of 2007 is the largest in the nation’s history. He also cautioned that education is not the same thing as skill, noting a study of information technology (IT) workers in the early 1990s that found that only a small fraction had degrees in IT fields.1 Noting that demands for specific skills and education were “fungible” to the extent that employers would provide specialized training, Cappelli asked Holzer which aspects of the future skill supply he was most concerned about.

Holzer replied that he is concerned about the share of workers with the broad competencies discussed earlier, not only the share with college degrees. Holzer noted that, because the largest group of immigrants has little education and few skills, they would increase the share of the national workforce that is not capable of moving into mid-level jobs. He emphasized that he does not consider this possibility a “catastrophe,” noting that it might draw attention to the need for policies to reduce inequality. Clarifying that he was not arguing that the national economy is “going to collapse because of the baby boomer retirements,” he nevertheless predicted that there would be increased tightness in certain markets for skill. Responding to another question, Holzer said that, if demand for skills keeps rising as it has in the recent past, the skill supply might not keep pace, for two reasons he had already mentioned—the immigrants making up a growing share of the labor force would have fewer college degrees than the retiring baby boomers they replaced and rising college costs would make it harder for lower to middle-income people born in the United States to obtain college degrees.

Holzer asserted that the continued rise in the wage gap between high school and college-educated workers provides evidence of growing skill demands, but Cappelli argued that this growing wage gap results from the “collapse” of wages for high school-educated workers in the early 1990s.


Responding to a question about immigration from Africa, Lowell said that there has been a large increase in migration of highly skilled Africans to the United States over the previous 10 to 20 years. He said that the level of “brain drain” from some of the smaller Sub-Saharan African nations is extremely high, and that some researchers are studying this phenomenon. Responding to another comment, Holzer said that some jobs being filled by immigrants—such as those in nursing—could be filled by native-born Americans if they had improved access to education and training. He noted that there is strong labor market demand for nurses, but that efforts to help more Americans obtain the credentials needed for this profession—such as by helping them enroll and graduate from four-year college programs—have not been as successful as hoped. Cappelli argued that many people are aware of the growing demand and want to enter nursing education programs, but the problem lies in constraints on these programs’ ability to grow.

David Finegold said he expected that, in the future, more science and technology graduates in China and India would choose to work in their home countries, rather than emigrate to the United States. Lowell responded that, although there might be some truth to this argument, the “population gap” between the developing world—the source of most skilled immigrants—and the developed world, including the United States, is “really astounding.” He argued that, despite the growing availability of higher education in STEM fields in India and China and the growing job markets there, many young STEM professionals will continue to emigrate to the United States. He said that wage differentials will continue to be the driving force in immigration, and that increasing the education of young people in other nations actually gives them a stronger incentive to bring their skills to the United States, where the rewards for such skills are higher. Lowell concluded that he could not foresee any future shortage of high-skilled immigrants to the United States.

Joining the discussion, Cappelli said that, although employers assert that they need more visas for foreign-born engineers to work in the United States, because not enough native-born students are studying engineering, the reason few young people born in the United States are studying engineering school is that the jobs do not pay as well as jobs in other fields. Bringing in more immigrant engineers would simply reduce wage levels further, Cappelli said. He suggested that, if the labor market adjusted to the current situation by increasing wage levels, that adjustment might be good for workers. Phrasing it as a question, he asked, “Is it a crisis if employers have to pay more for some of these skills?”

Herman “Bud” Meyers (University of Vermont) suggested that, rather than asking whether the policy community would take on the task of providing skilled labor to employers, Cappelli might rephrase the question to ask whether policy makers would partner with companies, encouraging them to invest in the education system. In Vermont, he said, property taxes are inadequate to fund education, and financial support from employers is essential. Cappelli responded that employers might choose to leave Vermont, locating in other states that do not seek such investments. He noted that North Carolina had attracted some large manufacturing firms by promising to provide all workforce training costs. He repeated his earlier point that the greatest change in employers’ behavior is their shift to outside hiring, rather than internal development of employees. He predicted this change would lead to more layoffs during downturns and more hiring when business is good, saying that policies would be needed to help workers with more frequent employment transitions.

Kalleberg concluded the session with several observations. First, he said that the panelists had reinforced his view that future skill demand could not be forecast with any accuracy. Second, he said that the number of workers appeared adequate to meet demand, but that there might be shortages of skills in certain sectors. Reflecting on the point that immigrants will provide the main source of employment growth in the coming decades, Kalleberg called for developing effective ways to teach immigrants teamwork, communication, and other broad skills. Finally, Kalleberg said, the discussion of skill supply and demand highlighted the potential value of policies aimed at encouraging employers to choose the “high road”2 by increasing workers’ skills, productivity, and wage levels.



An analysis of IT professionals in 1998 found that less than half had a bachelor’s degree with a major or minor in computer science or information science (Ellis, 2000).


Choosing the “high road” refers to reorganizing work, increasing worker training, and adopting other human resource practices that enhance skills, productivity, and wages (Appelbaum and Batt, 1993; see also Center on Wisconsin Strategy, 2007).

Copyright © 2008, National Academy of Sciences.
Bookshelf ID: NBK4071
PubReader format: click here to try


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...