Data Science in the Public Interest: Improving Government Performance in the Workforce. By Joshua D. Hawley. Kalamazoo, MI: W. E. Upjohn Institute for Employment Research, 2020, 152 pp., https://doi.org/10.17848/9780880996754.
Recent developments in web and mobile technologies have led to an explosion of big data, which are characterized by the so-called three Vs: volume (amount), variety (diversity), and veracity (accuracy). Because data can provide an efficient, fact-based means to identifying labor supply and demand, they can help governments allocate resources more effectively and prospective workers choose careers that are both profitable and beneficial to the overall economy. In Data Science in the Public Interest: Improving Government Performance in the Workforce, author Joshua D. Hawley examines how, by improving the quality, accuracy, and breadth of data within the public sector, data science can inform government workforce decisions. The author identifies several main workforce problems that guide the discussion.
One key problem is the long-term health of the workforce, which depends on the existence and maintenance of a mature workforce data system. In the 1960s and 1970s, the U.S. Department of Labor (DOL) and the U.S. Census Bureau developed longitudinal data systems to help federal and state governments make better decisions. Government researchers began to use these systems to study changes in employment and labor mobility, developing techniques for linking longitudinal data to decision making. Since the 1970s, however, many state governments have not updated their data systems. Hawley observes that some state governments still run their unemployment insurance programs on old computer systems that use outdated software written in Common Business-Oriented Language (COBOL), and there have been media reports that, during the coronavirus disease 2019 (COVID-19) pandemic, some jurisdictions recruited COBOL developers. As one example of a solution to this problem, the author frequently brings up Ohio’s longitudinal data system (developed jointly with DOL) as a paradigm of successful implementation of data analytics and machine learning in public policy. The system, which links information about K–12 and higher education with workforce development data, provides critical performance metrics that have helped Ohio launch effective responses to health and economic crises. These responses have ranged from analyzing and addressing surges in state unemployment to targeting food aid across local school districts.
Another workforce problem identified by Hawley is automation, which, according to one source cited in the book, will replace 73 million U.S. jobs by 2030. The fear of automation is widespread: workers, particularly those performing routine tasks, fear that they will lose their jobs; employers worry that they will struggle to find workers who can use modern technology; and governments are concerned about automation’s economic and fiscal impact on their communities. To address these concerns, governments have turned to big data for solutions. For example, states are increasingly using administrative data to design and build data systems that contain and track information on student outcomes, human services, workforce development, and employment. In 2014, the federal government outlined, for the first time, a common set of national performance standards in the Workforce Innovation and Opportunity Act (WIOA), and these standards have informed initiatives for providing and funding career and technical education, adult education, and temporary cash assistance for families in need. Some states and counties have incorporated WIOA performance standards into their economic development strategies, using online tools, such as dashboards, to monitor and measure the success of their workforce programs and systems.
A third problem discussed in the book revolves around the federal minimum wage, which has failed to keep up with increases in the cost of living. As seen during the 2007–09 Great Recession, which left many people unemployed, the problem of stagnating earnings was particularly stark among workers lacking the education or skills necessary to meet the demands of the changing nature of work. According to Hawley, one potential solution to this problem is government regulatory intervention through minimum-wage laws; another involves improving the skills of workers left on the sidelines by providing them with job training or subsidized postsecondary education. The author suggests that data systems can be particularly useful in informing these efforts, because they can help identify job skills that are in high demand. While the federal government has overseen vocational training for adults, states have shown renewed interest in funding specific workforce development programs, such as the Registered Apprenticeship Program and career and technical education programs for high school students.
Finally, Hawley turns to the problem of state governments being slow to learn from past successes and setbacks. According to the author, data on prior experiences with workforce development programs can be used to improve the performance of ongoing and future programs. One example given in the book involves Ohio’s Comprehensive Case Management and Employment program, which was rolled out in 2015 and required significant changes to the organization and delivery of local education and employment services to eligible youth. The state collected data on how counties changed their institutional structure to comply with state laws and implement the program, but it did not build the systems necessary to capture lessons learned from these changes. Thus, it became difficult to adjust program design, revise the curricula of the programs, or know whether program performance can be improved. Eventually, Ohio did build a dashboard using big data to create a set of standardized metrics for assessing program performance and presenting program outcomes. It is these lessons learned, captured in data, that could inform workforce development initiatives in the future.
While workforce data systems have continued to evolve since the 1980s, change has been slow. In 2010, in an effort to strengthen longitudinal data systems in state government, DOL launched the Workforce Data Quality Initiative, which has encouraged states to use data more actively in their workforce programs. While such initiatives are steps in the right direction, their success is not guaranteed. Convincing states to take the lead on developing workforce data systems and ensuring their interoperability may require a cultural change.