Photo by Alexey Demidov
Aaron R. Williams and Claire McKay Bowen
Deciphering the economy is difficult even under ideal circumstances. “Monetary policy is like driving while looking through the rearview mirror” is a popular expression in monetary policy, where policymakers make high-stakes decisions using data about the labor market, inflation, and economic growth. That’s because even the best data are released days, weeks, or months after they are collected. The record-long government shutdown has shattered that rearview mirror, leaving policymakers navigating a winding road without visibility, as the release of essential labor-market data remains disrupted.
The Labor Market is Confusing
Newspaper headlines have predicted eleven of the last three recessions.
Every day, we see headlines about job losses because of generative AI, the end of Covid-19 era employee hoarding, unpredictable tariff policy, and declining employment prospects for recent college graduates. Without federal data, our understanding of the labor market is overly dependent on anecdotes and high-profile layoffs from a handful of private companies. Without these essential federal economic data, we have limited capacity to sort through the confusion and make statements about America’s huge and complex labor market. We are shattering what little vision we had on the economy. Whether setting interest rates on the Federal Open Market Committee, managing a small business’s hiring decisions, or leading a local nonprofit trying to understand where to help, evidence-based decision making requires data to separate the signal from the noise.
Initial Unemployment Insurance Claims
Weekly initial unemployment insurance claims are a rare real-time measure of the labor market and are frequently the first indicator of trouble. Each week, the U.S. Department of Labor collects and processes claims data from each state’s unemployment insurance program office and then publishes these closely watched data.
The idea is simple: if employers start shedding jobs, the first place we’ll see economic trouble in a broad-based, quantifiable way is when newly unemployed people apply for benefits for the first time. For example, initial claims data guided policymakers through the pandemic and the post-pandemic inflation. Data released on March 14, 2020 showed about 252,000 initial unemployment insurance claims. One week later, initial claims increased to more than 2.9 million and peaked at over 6.1 million on April 4,2020.
The most recent release from the Department of Labor this year was on September 25th and covered the week ending September 20.
Friday Jobs Day Data
The first Friday of each month is every labor market wonk’s Super Bowl because the federal Bureau of Labor Statistics releases the Employment Situation Summary, or “jobs report.” Stories about this release end up on the cover of major newspapers, move financial markets, shape business and consumer sentiment, and inform political attitudes.
The jobs report is the product of two data collection efforts:
- The “establishment survey,” which uses responses to the Current Employment Statistics survey of businesses, estimates the total number of nonfarm jobs created and lost (+22,000 in August) and the number of jobs created and lost in major sectors (job gains in health care offset losses in the federal government and in mining, quarrying, and oil and gas extraction in August). The report also includes data on hours worked (the average workweek for all employees on private nonfarm payrolls was 34.2 hours in August) and earnings (average hourly earnings for all employees on private nonfarm payrolls rose by 10 cents in August).
- The “household survey,” which uses responses to the Current Population Survey, measures labor force status by demographics. In August, the unemployment rate remained steady at 4.3 percent and the number of unemployed people changed little at 7.4 million.
The most recent release from the Bureau of Labor Statistics was September 5 and covered data for August.
American Community Survey
The American Community Survey (ACS) is a premier source of information about small areas like counties, cities, and towns and detailed disaggregated data. The ACS is released less frequently than other data sources but provides much more detail due to its extensive questionnaire and large sample size. The 1-year ACS includes almost 2 million final interviews annually and supports a wide range of analyses, such as the prime-age employment rate at sub-state levels and unemployment rates by occupation.
For example, a state policymaker might want to examine unemployment rates and earnings for younger workers by educational attainment within their state to better understand the complex labor market for recent college graduates at their state institutions.
The U.S. Census Bureau releases tabulated data and microdata for the 1-year ACS annually. Fortunately, the Census Bureau released tabular 2024 ACS data in September 2025. Unfortunately, the 1-year ACS microdata, originally scheduled for release in October 2025, remain unavailable.
—
These are just three of many examples of how the U.S. federal government shutdown is limiting changemakers’ ability to respond to a dynamic world. Driving while looking through the rearview mirror for a short period of time isn’t too dangerous. Just keep the car pointed in the same direction. But the longer we drive while looking through a shattered rearview mirror, the less certain we can be that we’re still on the right path, and the more policy will err causing our communities and nation to suffer.
To raise awareness, we’re launching a series of blog posts and other outreach materials to help diverse audiences understand and support the federal statistical system. Everyone living in the United States is part of this vast statistical ecosystem and benefits from it—both directly and indirectly. Our first blog highlighted examples ranging from reducing lead exposure in consumer products to improving agricultural productivity.