How to Make the BLS Better

Good morning, Chairman Walberg, Ranking Member Sablan, and distinguished members of the House Committee on Education and the Workforce Subcommittee on Health, Employment, Labor, and Pensions. I am grateful for the invitation to discuss how the work done by the Bureau of Labor Statistics (BLS) serves policymakers, researchers, and ordinary people. I am a research fellow at the Mercatus Center at George Mason University, where my previous research has left me well acquainted with data provided by the BLS, particularly the Current Population Survey, which provides much of the information for the BLS’s monthly updates on the state of the economy.

The underlying goal of my testimony is to help members of Congress better understand the BLS and the information it provides. The main takeaways are as follows:

1. The BLS, now and throughout the past, represents some of the best economics research available. Its data collection and analytical work often set the gold standard that other economists attempt to emulate.
2. The weak link in the chain is in how the BLS communicates the information it has developed. The BLS website is the primary platform through which data users interact with the BLS, but despite efforts to make it more user friendly, attempting to access BLS data and understand it accurately can feel like wandering through a labyrinth. 
3. Leaders at the BLS recognize the agency’s struggles in this area and are taking steps to improve. Improvements should include communicating not only to professional economists, but also average Americans. The St. Louis Federal Reserve Bank’s work in this area might be a starting point.

The Origin of the BLS

The BLS has a more intriguing history than many government agencies. Its precursor, the Bureau of Labor, was created in 1884 amidst a surge in popular support for labor issues as workers’ bargaining power diminished with increasing industrialization. The mission of this agency was to “collect information on the subject of working people and the ‘means of promoting their material, social, intellectual, and moral prosperity.’”

Early labor advocates envisioned the Bureau of Labor as an inside-government agent through which to advance their agenda. Indeed, this had been the case in Massachusetts, the first state to create a bureau of labor statistics. The initial directors of that agency invited severe criticism by publishing research biased by their pro-labor views, and in response opponents called for the bureau to be abolished. Instead Governor Washburn appointed Col. Carroll Wright to head the agency, arguing that the correct response was “not in discontinuing the investigation . . . but in lifting it to a higher and broader level.”

Although Col. Wright was neither an economist nor a statistician, he brought a dedication to impartiality to the position and rebuilt the agency’s reputation to be focused on objectivity. He was so successful that President Chester A. Arthur, after a long struggle with labor advocates’ attempts to hijack the newly created federal Bureau of Labor, appointed Col. Wright as the first Bureau of Labor commissioner. Wright continued to serve after the agency gained department-level status, establishing a reputation for even-handed analysis that continues to this day.

Data Collected by the BLS

In keeping with the objective nature of BLS research, the modern mission statement is less labor focused, calling for the BLS to “collect, analyze, and disseminate essential economic information to support public and private decision making.” To this end, the BLS collects data on the following:

• Employment
• Unemployment
• Productivity
• Compensation
• Prices and Inflation
• Expenditures and Time Usage

BLS-sourced data is considered to be of the highest quality by most economists, and BLS survey methodology and subsequent data processing generally set the “best practice” standard that other researchers aspire to. The most-used BLS data collections include the following:

• Current Employment Statistics (CES)
  • State- and metro-level monthly data from employer surveys on employment, hours worked, and worker earnings by industry
• Current Population Survey (CPS)
  • State- and metro-level monthly data from household surveys on employment, hours, earnings, and information on special considerations such as union affiliation and contingent workers
• Job Openings and Labor Turnover Survey (JOLTS)
  • National monthly data on available jobs, new hires, and separations from employment
• Business Employment Dynamics (BDM)
  • Quarterly data reported by individual employers on the net change in employment at their establishments
• Quarterly Census of Employment and Wages (QCEW)
  • Industry-specific quarterly data on employment and earnings reported by employers as part of unemployment insurance program participation; available at the county level and covers nearly all (over 95 percent) of official jobs
• Occupational Employment Statistics (OES)
  • Annual data on employment and earnings for over 800 occupations available for metropolitan and nonmetropolitan areas
• National Longitudinal Surveys (NLS)
  • A set of long-running surveys tracking the employment, schooling, and social situation of a given group of individuals throughout their lives
• American Time Use Survey (ATUS)
  • Quarterly data from individual surveys available at the national level estimating the time spent by Americans in a variety of activities

Who Uses the BLS Data?

Although it can be argued that all Americans, to varying degrees, benefit from the data collected by the BLS, the primary users of the data are various researchers and media professionals who have the expertise to search out, engage, and digest the data. These primary users then deconstruct the raw data into usable information for policymakers and the public.

In its current state, the average person—a prospective employee, business owner, or student choosing a future career—would be hard pressed to find, analyze, and fully understand most of the BLS data. The complexity of the underlying methodology, specificity of information definitions, and difficulty in access and analysis mean that in its current state the BLS data requires a clerical class (commonly known as “wonks”) to interpret what seem like economic tea leaves.

The Weak Link in the Chain: Communicating the Data

The greatest obstacle facing the BLS seems to be communicating information clearly and accurately. Despite my own relatively extensive experience using BLS data, I still encounter difficulty in either finding what I need or even knowing about the existence of data that would benefit my research. In fairness to the BLS, this difficulty occurs in part because it is the curator of an enormous amount of complex information, and there is naturally a steep—and long—learning curve to fully understand what data are available and where they are kept. To modify a metaphor, Rome wasn’t built in a day, and it would probably take more than a week to read through the Library of Alexandria.

However, the tools through which BLS makes its data available don’t seem to make the task much easier. Searching the BLS website, for all of the effort to try to make it user friendly, often feels like wandering through a labyrinth. And upon finding the needed information, it’s often in a format that requires subsequent modification in order to be usable.

Perhaps most telling is the fact that intermediaries have emerged to offer researchers the raw BLS data in a more accessible fashion. For example, the Unicon Research Corporation previously offered the proprietary CPS Utilities software, which cleaned up the raw Current Population Survey data to allow for easier analysis. The Minnesota Population Center’s Integrated Public Use Microdata Series (IPUMS) has now taken over as the go-to source for easily accessible CPS data. Its website is substantially easier to navigate, and its presentation of the data allows it to be more readily usable and understandable.

The good news is that the BLS seems to recognize that its website and data handling software make the task of finding, understanding, and using its data more difficult. BLS administrators acknowledge the difficulties facing users and are doing what they can within the constraints of their current system to make it better. This suggests, though, that perhaps what the BLS needs is to build a better mousetrap from the ground up. The success of IPUMS and the Federal Reserve Economic Data (FRED) database maintained by the Federal Reserve Bank of St. Louis may offer some starting points for consideration.

Lastly, the BLS should consider expanding its engagement to ordinary people in addition to professional economists. For example, the primary definition for unemployment used by the BLS (the U-3 estimate) makes a great deal of sense to economists—it is accurate for what economists consider unemployment to be. But requiring that a person have actively looked for work within the last four weeks and be currently available to take a job (the required characteristics to be considered officially unemployed) seems arbitrary to noneconomists, leading to suspicion that policymakers are influencing official statistics in favor of one political party or another.

A better approach would appeal to both economists and John Q. Public. Adding additional measures of unemployment that make more intuitive sense to ordinary people, such as the Comprehensive Jobless Rate, could help the BLS engage with noneconomists and provide a starting point for ordinary people to understand the national economy better.

The Comprehensive Jobless Rate is the most holistic measure of unemployment. It counts all adults and adolescents who say they want a job as unemployed. In doing so it provides an upper bound on joblessness and therefore a comparison benchmark for the official BLS unemployment measures (it’s even useful to economists for this reason). 

The BLS has avoided similar unemployment measures in the past because of the “cheap talk” nature of such a response—it’s easy for survey respondents to say they want a job, whereas actively pursuing employment signals a “revealed preference” that a person truly wants to be part of the labor force. But ordinary Americans may know friends and family who are discouraged in their job search and have given up actively looking for work. The fact that the official unemployment measures fail to count these people as jobless ignites suspicions and distrust in the official estimates. The Comprehensive Jobless Rate could serve as a bridge to reengage with these people and therefore help them understand the overall economic situation better.

Conclusion

The BLS is rightfully regarded as an objective, data-focused organization whose efforts are essential to a better understanding of the US economy. Its data collection and analysis set the professional standard for many economists to follow. However, ordinary Americans would have great difficulty using the BLS’s data resources to answer their own questions about the economy.

This might be unavoidable—we shouldn’t necessarily expect that deep economic understanding is commonplace (in fact, some people might argue that deep economic understanding isn’t even commonplace among economists). Regardless, a worthwhile endeavor would be to make the data curated by the BLS more useful and more easily accessible to ordinary people, as well as to the economists who use them on a regular basis, in view of the important information the BLS provides.