January 2022. GrowthPolicy’s interviewed Joseph Fuller, Professor of Management Practice at Harvard Business School and Faculty Co-Chair of the “Managing the Future of Work” Project, on new workforce development models, the impact of AI on online job applications, and policy measures for a post-COVID labor market. | Click here for more interviews like this one.
Links: | | | |
Recent publications: | “” | “” (Harvard Gazette interview) |
GrowthPolicy: You are the Faculty Co-Chair of the Project on at Harvard Business School (HBS). Please tell our readers the origin story of this project—i.e., when and why it started, and what you hope it will achieve. Why do you believe we need new workforce-development models in the twenty-first century?
Joseph Fuller: After the great recession of 2008-2009, HBS launched a project on the competitiveness of the U.S. economy. The faculty overseeing the work had surveyed our alumni to assess America’s performance relative to our peer competitors on multiple dimensions. They identified the American workforce as a source of competitive advantage historically that had waned dramatically in recent years.
I knew from my work in industry that many businesses shared that point of view, so I decided to study the topic. Over time, the scope of our research expanded to include the skills gap, the impact of care-giving on workforce participation levels and diversity, credential inflation, and the impact of technology on the nature of work.
The school decided to launch a stand-alone project, Managing the Future of Work, and Professor Bill Kerr joined me as the co-head. We now have an eponymous in our MBA program and a that recently registered its 1 millionth download, as well as a student group and an executive education course.
GrowthPolicy: The COVID-19 pandemic has become a watershed moment redefining work and compensation. If you were to suggest some actionable policy measures, what would be the four or five most important ones to which policy makers should pay attention?
Joseph Fuller: Covid served as an accelerant for a number of trends that were latent in the labor market before the pandemic. Our skills development system—K through 12, post-secondary, and other training providers—simply isn't keeping up with changing skills requirements. The federal government should relax the conditions for the use of student loans to allow them to be used for career and technical training. Pell grant levels should be significantly increased, as well, and an equivalent mechanism to support life-long learning.
Perhaps most importantly, states should dramatically expand work-based learning opportunities for high schoolers. Our system is designed as if 100% of high schoolers will go on to a four-year college; roughly 60 % won't. Work-based learning programs allow teenagers to acquire technical skills and develop critical social skills necessary for making the transition from education to employment successfully. Participants in such programs have been shown to be more likely to complete high school and to pursue post-secondary education.
The federal government must also develop policies to reverse declines in workforce participation. Workforce participation in the U.S. peaked in the early 2000s and has drifted down ever since. That will require providing more widely available, affordable childcare. Companies should be incented to develop programs tailored to the needs of “hidden worker” group, ranging from the cognitively diverse to the formerly incarcerated.
We should also encourage more seniors to defer retirement if they’re so inclined. Data indicates that staying active yields better health outcomes. Continuing to work also reduces the rate at which retirement savings are drawn. Why not put an age cap on the collection of social-security payroll tax collections, giving older workers an automatic 6.2 % raise and employers an incentive to hire or retain older workers?
GrowthPolicy: Over the past decade we have moved to a “gig economy” in which part-time workers take on multiple jobs as independent contractors bereft of the safety net of health insurance or retirement benefits. As a scholar who studies the future of work, what do you believe policy makers should know about the long-term consequences of precarious labor and a contingent workforce?
Joseph Fuller: Many people hear “the gig economy” and immediately think of Uber and DoorDash. There are legitimate concerns about working conditions and access to benefits like health care and paid vacation for some categories of gig jobs.
But the gig economy consists of multiple, distinctive segments. For example, there has been dramatic growth in high-skilled, gig jobs, in fields like cyber security, digital marketing and cloud computing. Marketplaces like , , , and offer access to world-class talent to companies that simply can’t hire such workers on a full-time basis.
Moreover, a significant majority of contingent workers use gigs to supplement their income. Surveys conducted before COVID-19 consistently showed that on the order of 70% of such workers were happy with their gigs. So, policymakers should exercise care in regulating the gig economy, lest they undermine those parts of the market that work well for workers and employers. They should focus on defining what constitutes full-time employment and what protections such employees should enjoy.
GrowthPolicy: I’d like you to speak to the impact of Artificial Intelligence (AI) and automation on work. What role, if any, is technology playing in impacting real wages and exacerbating income inequality?
Joseph Fuller: AI is a core technology and will affect work in multiple ways. The most obvious is AI’s deployment to improve decision-making processes ranging from de-bottlenecking the supply chain to evaluating risk. Like most technologies, it has simultaneously displaced some workers, while creating fewer new jobs that pay well and have high skills requirements.
AI is used in the Applicant Tracking Systems (ATS) that are almost universally used to evaluate job applicants. My recent paper, , indicates that the AI-powered algorithms embedded within ATS screen out a large proportion of applicants for reasons other than their skills or experience. For example, the [ATS] systems routinely excluded candidates with a gap on their resumes of more than six months, on not having a college degree, or on having a criminal conviction. We estimate that over 25 million Americans are affected by this phenomenon.
People may be inclined to attribute all of that to bias. While instances of bias have been demonstrated by other scholars’ research, I believe it is actually a function of an excessive focus on efficiency. Companies program the AI to look for the closest possible fit with their idealized characterization of a candidate. They want to keep the number of candidates they consider to an absolute minimum in order to reduce the time and cost it takes to fill a position. Bizarrely, that focus on efficiency has the effect of eliminating a large number of potentially qualified candidates.