From Agencies to Agency: Building a Workforce from Within

By Brent Orrell

In his 1980 presidential campaign, Ronald Reagan popularized the importance of work as the primary means of self-sufficiency by telling audiences, “The best social program is a job.”1 In 2015, President Barack Obama echoed the same idea, saying, “The best antipoverty program is a job.”2 As a matter of national character, Americans of all backgrounds and beliefs agree: The key to self-sufficiency and intergenerational wealth accumulation is to ensure that every person has the education, skills, and opportunities they need for gainful, satisfying employment.

Key Points:

  • Existing programs leave many disadvantaged Americans facing technological developments and globalization without gainful and meaningful employment.
  • Reinvigorating the workforce will require programs that emphasize personal agency, improve workers’ understanding of labor market information, and empower workers to cultivate interests, knowledge, skills, and abilities.
  • Policymakers should curb poverty and shore up agency by incentivizing soft skills development, improving labor market information systems, and equipping workers with flexible Individual Training Accounts.

A shared belief by itself, however, cannot guarantee the outcomes we seek for ourselves and our communities. Technological change, the globalization of the economy, and stagnating educational performance and work readiness have conspired to put “good jobs at good wages” out of reach for a growing number of Americans, especially those who live on the social and economic periphery.

Declining access to good employment is a multidimensional problem that will require solutions tailored to the specific needs of local and regional economies, businesses, and—of greatest importance—workers themselves. Our existing workforce development system is not up to this challenge. As my AEI colleague and workforce systems expert Mason Bishop says, “We have a New Deal workforce system for an iPhone economy.”3

The modern economy is “dematerializing” as our processes become more efficient and we advance further into job markets dominated by services and information-related skills.4 The result of this shift is a market premium for noncognitive or “soft” skills—such as communication, teamwork, critical thinking, and grit—that are challenging the dominance of the types of narrow, repetitive technical skills that were more common in the industrial age.5 Correspondingly, the systems and approaches we’ve used to assist workers relying on cumbersome, in-person service provision, often outdated technology, and a two-steps-behind understanding of market changes are inadequate to the needs of workers scrambling to meet the demands of the economic moment.

In this context of rapid change and relatively static systems, it is also clear that American workers have not been keeping up. In their seminal book on this topic, The Race Between Education and Technology, Harvard professors Claudia Goldin and Lawrence Katz argue that the virtuous cycle of rising educational attainment, jobs, and income that characterized the American economy from the dawn of the Industrial Revolution until the 1970s has broken down.6 The pace of skill acquisition among American workers has simply not kept up with technological change, and, as a result, family-sustaining jobs have gradually moved further out of reach for growing numbers of workers. With the advent of powerful new tools such as artificial intelligence and robotic process automation, the task of synchronizing education, training, and technology is becoming more difficult, not less—especially for those with lower levels of education, training, and skills.

As I wrote in a 2018 report, the roots of our work-readiness and job-training challenges are deep and have their ultimate source in family formation and the problems that underprepared and under-supported children have in maximizing school performance.7 Our long-term workforce problems are, in this view, part of a broader human development problem. Other contributors to this volume are addressing these underlying challenges, and I will not rehearse their scholarship here.

Rather, in this chapter, I focus on the question of how we can better deploy and align our fiscal resources, workforce systems, and programs to help new and incumbent workers start or restart their careers. To help workers, especially those in low-income and disadvantaged communities who are the special concern of this volume, my recommendations focus on four areas:

  1. Improving noncognitive (social-emotional) skill development through increased investment in evidence-based programs such as home-visiting programs that strengthen educational readiness, long-term educational attainment, and employment outcomes;8
  2. Shifting workforce development resources from bureaucratic systems to individuals through increased use of Individual Training Accounts (ITAs) that build work skills and reinforce the development of personal agency;
  3. Reforming, updating, and enhancing labor market information technology and systems to improve worker awareness of in-demand jobs and skills; and
  4. Promoting state-level flexibility through federal waivers that promote innovation in workforce system redesign.

Improving Noncognitive and Soft-Skills Deficits

Over the past 70 years, Americans have invested heavily and disproportionately in so-called hard, cognitive, or technical skills in the hope and belief that these skills are the surest path toward the economic security that high-paying careers in STEM fields afford. Yet the skills gap between what employers say they want and what American workers have to offer continues to grow.

Workforce success is built on a pyramid of skills. Figure 1 is a rendering of that pyramid. At the top of the pyramid are technical skills that focus on either particular tools or processes that are most directly applicable to the job market, especially the digital skills that are increasingly crucial across the economy. In the middle are basic skills such as reading, writing, and mathematics—core work competencies for literate and trainable employees.

Figure 1. The Workforce Skills Pyramid

Source: Adapted from CareerOneStop, Competency Model Clearinghouse, “Building Blocks Model,” 2018, See also Brent Orrell, ed., Minding Our Workforce: The Role of Noncognitive Skills in Career Success, American Enterprise Institute, May 18, 2021,

The base of the workforce pyramid is noncognitive or soft skills such as communication, critical thinking, teamwork, grit or persistence, and collaboration. Workers who manage to land jobs on the strength of basic and technical skills but without strong noncognitive skills often find themselves disadvantaged in retention and advancement. As workforce development professionals say, “Technical skills get you hired, but noncognitive skills get you fired.”9 And even if they don’t get a worker fired, the lack of noncognitive skills tends to hamper progression toward higher-paying management jobs.

Employers confirm the need for a workforce with higher levels of noncognitive skills. Survey data from employers on LinkedIn show that industry experts, organizational partners, executives, talent developers, and managers all say soft-skill training was their top training priority in 2020.10 Talent developers in all sectors allocate the majority of their budgets toward soft-skill development programs rather than technical training. LinkedIn’s 2018 survey among these business and organization leaders best sums up the rationale behind soft-skill development and why it is a top priority today: “In the age of automation, adaptability rules. While maintaining technical fluency will be important, demand for soft skills will continue to accelerate.”11

Data from a recent AEI survey of STEM workers help explain why such skills are important over a career lifetime. AEI found that “69 percent [of] STEM workers say good written or communications skills are extremely important.” Forty-six percent said interpersonal skills are important, compared with only 36 percent who say “high-level math, analytical or computer skills” are important to do their jobs. The perceived need for such skills rises with age. As Figure 2 suggests, older STEM workers (the ones most likely to have advanced into management occupations) rate skills such as critical thinking, written and spoken communication, and interpersonal skills as significantly more important than their younger colleagues do.12 These data suggest noncognitive skills are crucial to turning a job into a career.

Figure 2. Skills STEM Workers Say Are Important to Do Their Job, by Age Group

Source: AEI STEM Education and Workforce Survey, 2019. See also Brent Orrell and Daniel A. Cox, STEM Perspectives: Attitudes, Opportunities, and Barriers in America’s STEM Workforce, American Enterprise Institute, July 15, 2020,

If noncognitive skills are so important, what can we do to get more of them? I covered this question in-depth as part of my 2018 report “STEM Without Fruit”13 and in a follow-up volume released in 2021, Minding Our Workforce: The Role of Noncognitive Skills in Career Success.14 In brief, the report and the volume argue we should be focusing on a life-cycle approach to workforce development to ensure that more kids arrive at school age better prepared to successfully complete K–12 education and advance into postsecondary training. This means greater investment in family stability and relationship-education programs and ensuring that our schools and community organizations, especially those that disproportionately serve low-income and disadvantaged students and communities, are better informed about the long-term impact of building these skills.

For older students preparing for careers, programs such as the Federation for Advanced Manufacturing Education, which uses an advanced manufacturing training approach, have shown great success at combining technical and noncognitive skills training and building pathways to long-term employment.15 For welfare-dependent families, programs such as Economic Mobility Pathways help participants improve goal-setting, decision-making, and problem-solving skills that help them reach and sustain economic self-sufficiency.16 For those already in the workforce, several highly successful model programs such as Year Up and Per Scholas focus on integrating technical and noncognitive skills training, resulting in rapid and dramatic increases in wages and workforce success.17

Increased Investment in ITAs

The previous section dealt with the need to strengthen America’s human capital base by placing more emphasis on developing noncognitive skills that undergird workforce success. Noncognitive skills training is meant to increase flexibility and longevity in the workforce by helping people gain the ability to relate well to others on the job and in the rest of their lives. For both practical and philosophical reasons, the workforce system itself needs to mirror and support this bias toward cultivating personal agency in training and career decisions.

Shifting more decision responsibility to workers is especially important in light of how quickly technology and skill demands are changing. No government agency can, by itself, fully and accurately predict what skills might be needed, especially on a short time horizon. Workers who are actively pursuing jobs may be in the best position to know what kind of training is most relevant to the market and their own interests. Putting resources in the hands of workers, therefore, can add to the agility and flexibility of our training system.

ITAs are one mechanism for achieving this objective. Replacing the 1998 Workforce Investment Act (WIA), the 2014 Workforce Innovation and Opportunity Act (WIOA) introduced these accounts as an effort to better serve the integration of workforce development and economic development. Rather than rely on local workforce system authorities to decide which training programs would receive public contracts, WIA authorized ITAs to empower workers in the same way a voucher is sometimes used in public school education. ITAs are used by workers to pay for training from a list of approved training organizations and entities including community colleges, technical schools, and private, for-profit training organizations.

Because this approach has been around for a considerable period, there has been time to test and evaluate its effectiveness. A government-funded, randomized control trial evaluation of ITAs conducted by Mathematica Policy Research (MPR) tested three different ITA models: structured choice (intensive case manager involvement), guided choice (moderate case manager involvement), and maximum choice (voluntary case manager involvement).

Per Table 1, MPR’s evaluation showed that Model 1 (structured choice) participants earned about $500 more per quarter than Model 2 (guided choice) participants did. Model 1 and Model 3 (maximum choice) were roughly comparable in outcomes. Over 20 years, MPR estimated that workers using the structured choice model would show “large, positive impacts on long-term earnings . . . while generat[ing] $46,600 per ITA job seeker in net social benefits.” The other models also showed significant income gains for workers using ITAs.18

Table 1. The Three Service Delivery Models Tested in the ITA Experiment

Source: Irma Perez-Johnson, Quinn Moore, and Robert Santillano, Improving the Effectiveness of Individual Training Accounts: Long-Term Findings from an Experimental Evaluation of Three Service Delivery Models, Mathematica Policy Research, October 2011, See also Brent Orrell, Mason Bishop, and John Hawkins, A Road Map to Reemployment in the COVID-19 Economy: Empowering Workers, Employers, and States, American Enterprise Institute, July 24, 2020,

The expanded use of ITAs is of particular relevance in the post-COVID-19 economic environment. Some Americans, especially those at the lower end of the wage scale, have been relatively slow to return to work for various reasons, including fear of illness, day care and school closures, and the availability of expanded federal unemployment benefits.19 The workforce has also been marked by historically high levels of quits, especially among low-wage workers seeking better pay and benefits.20 This labor market churn strongly suggests that many workers are anxious for advancement and motivated to seek the necessary skills to move up. For such workers, ITAs could be a vital resource for accessing the skills training they need to achieve higher earnings and better working conditions.

The stresses of the COVID-19 return to work provide a second reason for expanding individual choice and resourcing through ITAs. A survey by Strada Education Network found that over 30 percent of Americans would need more education or training to get the job they really want,21 a number that would swamp the capacity of the workforce system. Expanding access to ITAs would permit many of those workers to quickly access training resources while preserving limited case management capacity for workers who need more support and guidance.

Consistent with the principle of choice and to support the development of personal agency, Congress and the Biden administration should consider a significant expansion of ITA funding and some reforms to the use of these accounts that would make them even more flexible.22 First, Congress should pass legislation to increase funding for ITAs and direct decisions for the disbursement of those funds to state governors rather than sending the money directly to local workforce investment boards by formula. Under recovery conditions, governors are better positioned to understand overall state needs and ensure that regional and interstate-state economic clusters suffering from significant staffing shortfalls receive the resources they need to address the labor and skill shortages that will be of greatest benefit.

A second important reform would be to loosen restrictions on training programs that are eligible to receive ITA funding. Expanding options for training in ITAs would allow for a greater range of choices workers could use that best fit their needs and career interests. The accounts could be used to pay for traditional classroom-style learning at a community college, vocational training at a private or for-profit institution, on-the-job training as a stand-alone opportunity, or part of a certified or industry-recognized apprenticeship program.

Finally, restrictions should be relaxed on the types of services ITA resources can be used for. The top barriers to employment normally fall into one of three categories: housing, childcare, and transportation. At a minimum, workers should be able to apply some of their ITA to address nonwork barriers that are preventing the return to work. Individuals who face limited employment opportunities could also use these funds to offset relocation costs within and between states to move to an available job.

Improving the Quality and Accessibility of Labor Market Information

So far, we’ve examined the importance of noncognitive skills and argued that our workforce system should expand the use of ITAs, a policy vehicle that reinforces the exercise of individual choice and self-direction among workers and assists motivated workers to quickly gain skills they need to advance. To facilitate and support effective choices, however, workers and employers still need timely, accurate, and tailored information about the types of jobs available and the sorts of training and skills required to prepare for them. This is especially true for new or less-experienced workers, who often struggle to align their skills and interests to a regional or local economy. Labor market information (LMI) provides both context and direction for identifying, leveraging, and applying skills to the workforce.

In the past, workforce development professionals and labor market economists have collected LMI for aggregate-level analysis of labor market trends. While understanding these trends is extremely important for policymakers and analysts, LMI should also be made more accessible and approachable for job seekers, students, workers, and employers to inform their touch points with the labor market. Improving the collection, display, and accessibility of LMI would help workers, educators, and employers identify the range of possible opportunities and align their training programs to them.

The advent of powerful new information management and analysis tools has dramatically accelerated innovation in LMI. This has been complemented by large public and private investment in creating systems for making such information useful to Americans. The following examples comprise just a few of the efforts in progress around the country targeted at better integrating LMI and improving understanding among policymakers, workforce practitioners, and workers about the full range of relevant LMI and how to better use it.

  • The US Labor Department Workforce Data Quality Initiative. The initiative has invested nearly $70 million in competitive grants to help states modernize state LMI and educational data systems.23 At the height of the early COVID-19 pandemic, Congress provided $1 billion to the states to upgrade administration and technology for the nation’s unemployment insurance system.24 The improvements made through this spending need to be considered as part of broader LMI integration initiatives.
  • The Bill & Melinda Gates Foundation. The foundation is funding several multistate collaboratives to improve quality and interoperability of LMI systems. The National Association of State Workforce Agencies was a recipient of such funding for its National Labor Exchange data-matching and analysis project that will help the public sector get access to higher-quality and timelier LMI.25
  • The US Chamber of Commerce’s Jobs and Employment Data Exchange. The initiative addresses the issues of data standardization across jobs, skills, and credentials and aims to create a trusted repository of employment data from its member-employers nationwide.26 When the initiative is completed in 2024, the breadth and timeliness of the information will improve strategy and practices for employer-based LMI reporting for government evaluation and social science research.
  • The Alabama Terminal for Linking and Analyzing Statistics. This data-matching initiative aims to create a federated system of education and labor-agency databases through shared data naming and tagging.27
  • Atlanta United Negro College Fund. The fund is working with Emsi to standardize occupational language around skills with a new tool that identifies regional skills demand and helps workers build skill-based resumes.28
  • The Markle Foundation’s Skillful Initiative. This initiative is attempting to serve both state governments and businesses in transitioning job descriptions in the labor marketplace into skill-based nomenclature.29 Since skills are the DNA of the labor market and the smallest units in labor market analysis, clustering skills together and classifying them into a progression helps precisely and accurately define roles based on their essential components across a range of potential job opportunities.30

The databases described above create the foundation for more comprehensive and integrated LMI systems. These data still have to be shaped for use in public-facing tools available to workers and employers that are usually not experts in understanding LMI. The following are some examples of efforts trying to realize those next-step applications.

  • Kentucky’s Local Area Workforce Dashboard. The dashboard gives the public a way to see training and employment services availability, workforce program information, and other labor market information.31
  • The Kentucky Center for Statistics. The center has developed an online tool that offers several methods of exploring Kentucky-based jobs. Using an intuitive computer interface, users can complete a simple knowledge, skills, and abilities self-assessment that can help workers align skills, interests, and education with available jobs.32
  • The Alabama College and Career Exploration Tool. This tool is Alabama’s effort in addressing the need for LMI-augmented career-exploration resources and learning and employment records.33 This will function as a secure online portal for Alabamians that houses their education and employment records in a verified digital resume. It will also feature an Alabama-specific job and education marketplace and an education and career exploration tool based on county-level LMI data.
  • The New Jersey Department of Labor and Workforce Development. The department has developed an LMI resource on its website, featuring a Microsoft Power BI dashboard of relevant and malleable labor market indicators, an occupation-exploration tool, and a database of employers in the state.34

The above examples represent a small sample of the kinds of public and private innovations states and regions are pursuing to improve LMI accessibility and application. This means that one of the most helpful things the federal government can do at this point is serve as an investigator and researcher analyzing and evaluating the technology and practices under development. The US Department of Labor should invest in these monitoring activities to identify promising approaches and technologies and promote cross-fertilization between initiatives rather than prescribing a single, national approach to improving LMI systems.

Federal Waivers for Workforce System Innovation and Redesign

The discussion above relating to the problems in our existing approaches and systems to labor market data points to a much broader, long-term need: full-scale reform of how we organize and administer workforce development and the many overlapping programs that form the patchwork of supports that low-income and disadvantaged workers rely on. The bottom line is this: Just as we apply principles of choice and self-direction to workers in making decisions about training, jobs, and careers, we should apply those same concepts to the organization and administration of the workforce system itself to foster a greater sense of local ownership of workforce challenges and programs.

In 2019, Bishop wrote a report looking at the workforce-program landscape that brings into sharp focus just how complex and inefficient these programs frequently are.35 The point of Bishop’s observations is not so much an indictment of the way workforce programs have evolved over nearly 100 years but to point out how the complexity of the system itself erects barriers to self-efficacy and self-sufficiency. He points out three primary problems:

  1. Ad hoc program development with competing interest group constituencies and overlapping bureaucracies,
  2. Intrusive federal requirements that have inhibited reform and innovation at the state level, and
  3. Administrative complexity arising from steadily multiplying education, workforce, and social services initiatives and programs that leads to wasteful and duplicative systems at the state and local level.

As part of Bishop’s work, AEI produced a webpage that illustrates the patchwork of patchworks that is today’s workforce system, with each state organizing its programs in somewhat different fashions.36 The rule of thumb in looking at the individual state system structures is the more boxes at the lower level of the schematic, the more fragmented and duplicative the system is. Figure 3 shows a relatively complex organizational structure in one state (Alabama) and a streamlined and consolidated system in another (Utah).

Figure 3. Comparing the Public Workforce Systems of Alabama and Utah

Source: Information for infographics derived from federal and state agency websites from March 2019 to January 2020, alongside Workforce Innovation and Opportunity Act state plans. See also American Enterprise Institute, “Employment and Job-Training Reform: A Framework for Policy and Practice,”

Under a federal system that seeks to devolve power, resources, and administration away from the federal government, we should expect and welcome variation as states adjust programs to their own social and economic realities. But the complexity of many state systems is also the consequence of the way federal laws “silo” funding with a variety of program requirements, desired outcomes, and performance standards and restrictions that inhibit states trying to rationalize and connect programs serving similar and, in some cases, identical customers. These duplicative and competing programs drain resources away from people in need and toward systems that administer services (e.g., duplicative staffing and offices and competing data management systems). Finally, complex and overlapping structures create a bureaucratic thicket that is confusing and exhausting to navigate for administrators and citizens alike.

The good news is that some states, such as Utah, are on the way to taming the beast of bureaucracy in the workforce and social services and refocusing available resources on disadvantaged citizens’ needs. Beginning in 1992 as part of its welfare reform initiative, Utah used demonstration authority under Section 1115 of the Social Security Act to begin reorganizing and consolidating fragmented welfare and human services programs. Over the intervening three decades, the state has gradually corralled 26 federally funded programs into just two agencies: the Department of Workforce Services and the State Board of Education. Under the WIA, the state also received permission from the US Department of Labor to create a single workforce area covering the entire state, helping reduce inefficiency and maintain greater consistency of service provision across the state.37

Today, Utah has consolidated workforce development with Medicaid eligibility, housing assistance, refugee resettlement, and vocational rehabilitation programs to create a unique “one-door” system that helps Utahans access the full range of workforce and social and human resources programs they need to move toward self-sufficiency. The consolidated system also conserves resources and shifts more dollars toward services and away from administrative overhead.

The additional virtue of Utah’s administrative structure is that it has vastly simplified accounting for federal dollars and the state’s reporting relationship with the federal government. Department of Workforce Services case managers and other employees are randomly sampled regularly to keep track of how these workers are spending their time across the various programs. Using this sample, the state then reports its expenditures to the federal government under a cost-allocation formula rather than submitting detailed justifications for individual programs. Further, rather than reporting to multiple federal agencies, Utah’s consolidated workforce–human services system works exclusively through the US Department of Health and Human Services, which provides fiscal oversight and coordination with other federal agencies that contribute to Utah’s overall federal funding.

One response to the Utah experience would be to try to impose it on all other states. This would be a mistake. Utah is a relatively small state in terms of population, and what works for Utah might not work elsewhere. Federal social engineering and the overriding of the prerogatives and interests of state governments are also inconsistent with federalism generally. Rather, the lesson to be drawn from Utah’s experience is how it might be leveraged as a model for incentivizing and supporting state innovation itself.

This approach should sound familiar because it is similar to the path followed on welfare reform. Beginning in the 1980s, the federal government encouraged states to exercise Section 1115 waivers to improve welfare programs. States moved to test innovative methods for encouraging and requiring work and work-related activity in exchange for welfare benefits.

After a decade of experimentation, a Republican-led Congress and the Clinton administration recognized the success of these and other efforts and moved to transform the existing Aid to Families with Dependent Children program into the time-limited, work-focused Temporary Assistance to Needy Families (TANF) program. The TANF program continues today as a flexible block grant that allows states to tailor their welfare programs to low-income families’ needs and opportunities. Welfare caseloads have dropped dramatically, while incomes have gone up and poverty among children has gone down.38

Section 1115 authority, which dates to the Kennedy administration, remains in place and could be directed toward new state-directed efforts to reorganize, consolidate, and integrate workforce development and human services. As part of the next reauthorization of the WIOA, or separately, Congress and the Biden administration might explicitly encourage (not require) states to submit plans to the US Departments of Labor and Health and Human Services for their own service-integration plans to simplify and integrate their workforce, social and human services, and educational programs to better support economic self-sufficiency and harmonize program structures and administration. Since WIOA itself does not fall under Section 1115, additional flexibility from Congress to explicitly incorporate WIOA programs for this demonstration would be extremely helpful. Chapter 8 in this volume by my AEI colleagues Angela Rachidi, Matt Weidinger, and Scott Winship proposes a similar approach but recommends giving states the authority to reorganize and consolidate the balance of the safety net, including Supplemental Nutrition Assistance Program, Supplemental Security Income, TANF, and other safety-net programs.


Intentions and philosophy are not enough when boosting skills and employment among disadvantaged or low-income Americans. Work, and the dignity and self-sufficiency it fosters, is an indispensable good for individuals and society. Our challenge is to overcome mismatches between available workers and available work and ensure that every American has the education, training, and support they need to find, retain, and advance in employment.

For disadvantaged individuals, this means strengthening interventions that address both technical and noncognitive skill deficits, expanding programs that increase personal agency in training and employment, and improving labor market information to help individuals match their interests, skills, and training to employer needs.

Finally, creating a more robust and effective workforce system requires greater innovation in program design, implementation, and administration. While we cannot and should not reinvent the nation’s workforce system overnight, multiyear experiments by governors, state legislatures, and WIOA system administrators are needed to develop new strategies to boost program efficacy and outcomes. Our states are laboratories of democracy and economic development, and our policies should leverage and support them as assets to improving opportunity for all Americans.


Thank you to Jake Easter for his excellent research support.