Producing Gold-Standard Statistics: How the Bureau of Labor Statistics Should Adapt Its Unemployment Rate Metrics in Response to the COVID-19 Pandemic

Comment on the Draft Department of Labor (DOL) FY 2022–2026 Strategic Plan and Evidence-Building Plan

Agency: US Department of Labor, Bureau of Labor Statistics

Comment Period Opens: July 16, 2021

Comment Period Closes: August 6, 2021

Comment Submitted: August 6, 2021

Request for Comments: https://www.dol.gov/agencies/oasam/centers-offices/performance-management
-center/reports/FY2022-2026-strategic-plan

The US Department of Labor (DOL) has requested input on the drafts of its strategic plan and its evidence-building plan. I am a research fellow at the Mercatus Center at George Mason University, which is dedicated to advancing knowledge about the effects of government policy on society, including policies and programs regarding government-collected data. Accordingly, my comment seeks to aid the DOL—and the Bureau of Labor Statistics (BLS) in particular—as it considers how to adapt its operations to best fulfill its mission.

Introduction

The COVID-19 pandemic abruptly derailed the longest period of sustained economic growth in the United States. The resulting economic decline was more rapid and substantially different from previous recessions. A vivid symptom of this divergence became apparent when previously reliable labor market metrics lost explanatory power. The BLS has done an admirable job communicating how workers’ and businesses’ responses to the pandemic were affecting its data collection. However, almost a year and a half later there are still problems comparing prepandemic economic data with current economic data.

These problems go beyond the BLS’s identification of likely misclassification errors in the Current Population Survey (CPS). For example, the primary unemployment metric, U-3, declined from a peak of 14.7 percent in April 2020 to 5.9 percent in June 2021. After accounting for the likely misclassification errors, these estimates could be as high as 19.7 percent and 6.1 percent, respectively. But these data points aren’t comparable to the equivalent prepandemic unemployment rates because the makeup of the labor force substantially shifted in response to the health concerns of workers, businesses, and customers and in response to government social and economic restrictions and government relief programs. In fact, my estimate of the “prepandemic comparable unemployment rate” suggests that better estimates of U-3 for April 2020 and June 2021 might be 23.5 percent and 8.0 percent.

Furthermore, the increasingly acrimonious political environment over the past few decades has led to BLS data releases being used by pundits—and more recently, politicians themselves—as evidence in favor of their preferred policies or as allegations that the BLS has been captured by partisan interests.

The BLS remains a highly professional and nonpartisan institution. However, an effort to reengage the constituencies the agency serves could affirm both its intellectual authority and its commitment to the public interest. Reevaluating the existing labor market metrics and experimenting with new labor market metrics that better address the intuition, experience, and data needs of workers, businesses, and researchers would assist this endeavor.

Potential New Unemployment Metrics

I next outline a few novel unemployment metrics that have been proposed to measure slack labor in the workforce. Unemployment is arguably the most important place to start because those metrics are the most criticized and weaponized by political interests.

There are a number of reasons that the current array of unemployment metrics might not tell the full story. First, they were primarily chosen by economists for economists. The primary unemployment rate (U-3) is designed to measure the degree to which unemployed workers are affecting the market for labor, which is an important economic question, but that issue has little relation to the public perception of what constitutes unemployment. Similarly, the expansions to U-3—the discouraged worker unemployment rate (U-4), the marginally attached unemployment rate (U-5), and the underemployment rate (U-6)—attempt to understand how economic and social constraints reduce the supply of labor, but they provide indistinct, and sometimes even misinterpreted, economic insight.

More precisely tailored unemployment metrics, with clear explanations of their meaning, would improve public confidence in the BLS while also providing economists with better identification for quantitative research. The rest of this comment provides a range of examples for consideration.

Prepandemic Comparable Unemployment Rate (U-3_PPC)

The prepandemic comparable unemployment rate (U-3_PPC) offers a way to more realistically compare the current national labor market with the near-full-employment labor market that existed before the COVID-19 pandemic.There are two potential versions of the metric, one that takes a static approach and one that takes a dynamic approach. The static approach simply holds the size of the labor market constant at the February 2020 level, assuming that any nonworking individuals from that group should be realistically counted as unemployed. The dynamic approach starts from the same basis but attempts to incorporate the structural changes in the labor market caused by COVID-19 by accounting for the increased rate of retirements and number of younger workers entering the workforce for the first time.

Comprehensive Jobless Rate (U-CJR)

The comprehensive jobless rate (U-CJR) is a complement to the U-5 rate. It is arguably the highest feasible measure of joblessness that could exist, providing an upper bound for political arguments over how high the unemployment rate really is. It uses the standard construction of the labor market and simply counts everyone who says they want a job (and is not currently employed) as being unemployed. These counts include “cheap talk” survey responses from individuals who might not actually take a job if offered one (a concern that caused a similar unemployment metric to be excluded during prior unemployment metric reforms).

Comprehensive Underemployment Rate (U-CUR)

My own research in progress suggests a revision to the U-6 measure of underemployment. U-6 assigns the same weight to workers who are completely jobless as those who are already employed but would like to work more. Therefore, U-6 overestimates the true measurement of slack labor. The comprehensive underemployment rate (U-CUR) is constructed similarly to U-6 but weights each underemployed survey respondent by the degree to which they are underemployed, providing a more accurate measurement.

COVID-19 Underutilization Rate (U-Cov)

Jason Faberman and Aastha Rajan at the Federal Reserve Bank of Chicago have suggested another pandemic-focused alternative to U-6, which they call the COVID-19 underutilization rate (U-Cov). It adds the workers who are likely to be misclassified to the U-6 rate, as well as workers who say they want a job but are not counted in official unemployment statistics (similar to U-CJR). However, it keeps the same equal weighting of complete joblessness and underemployment seen in U-6.

Nonemployed-to-Population Ratio (NPOP)

Ernie Tedeschi, a former Treasury economist now working with the Council of Economic Advisers (CEA), has suggested the broadest possible estimate of unemployment in order to capture the full spectrum of economic effects attributable to COVID-19. The nonemployed-to-population ratio (NPOP) is the complement to the more widely used employment-to-population ratio (EPOP), measuring the share of the total population that is not employed full time or voluntarily part time. Tedeschi suggests that prime-age NPOP and a demographically adjusted NPOP—which controls for sex- and age-related changes in the composition of the population—would be necessary for accurate comparisons over longer time periods.

Population Hiring Rate (PHR)

Former CEA chair and Stanford professor Ed Lazear in 2015 proposed the population hiring rate (PHR) as a better—and more direct—measure of labor demand during the recovery from the global financial crisis. The PHR is the ratio of monthly hires from the BLS’s Job Openings and Labor Turnover Survey relative to the working-age population. Like Tedeschi with NPOP, Lazear suggests that PHR be modified by changes in the population composition to ensure accurate comparisons over time.

True Rate of Unemployment (U-TRU)

This list of alternative labor market metrics is not complete—other ideas may also be worthy considerations. But the BLS should exercise caution against partisan attempts to develop alternatives that benefit a particular political agenda.

I would argue that the true rate of unemployment (U-TRU), developed by Gene Ludwig, former comptroller of currency, is one such metric to beware. U-TRU purports to measure the share of the labor force that is “functionally unemployed.” Ludwig defines functionally unemployed individuals as those who have no job but want one (similar to U-CJR) or who are involuntarily underemployed (similar to the equal weighting of underemployment and joblessness contained in U-6) and, most problematically, who earn an annual pretax income of less than $20,000.

Although workers’ earnings are indeed important, conflating the concept of a living wage (which obviously is dependent on region and household, not to mention a highly subjective topic) with the concept of unemployment seems designed to create an artificially large unemployment metric that would undermine the BLS’s authority and facilitate political angst.

Conclusion

Better engagement with the public is critical to restoring BLS’s prominence as the most authoritative source of labor market statistics. The BLS’s current unemployment metrics are valuable but insufficient. A wide range of potential improvements could be made with the data collected from existing BLS surveys, as described in this comment.

The BLS could affirm its independence, nonpartisanship, and intellectual authority regarding labor market statistics by fast-tracking the development of new labor market metrics that more intuitively tell the experience of workers and businesses while continuing to provide researchers with critical data to analyze the economy.