The tale of the disappearing jobs numbers

E.J. Antoni, Tribune New Service

The nation’s highly anticipated monthly job reports have turned into the boy who cried wolf. Ever since the pandemic, these labour market estimates have been wildly inaccurate and required significant revisions. That’s troubling because major decision-makers from Washington to Wall Street no longer have reliable data, and the consequences affect every American family. The source of the problem is chiefly the Bureau of Labour Statistics (BLS), which produces monthly estimates on labor market conditions, like the unemployment rate and the number of nonfarm payrolls. They also produce the widely watched inflation statistics, the consumer price index.

In fairness to the BLS, measuring the economy, including the labour market, has been abnormally difficult since the pandemic because so much has changed since 2019. The work-from-home phenomenon is one example, but there have also been drastic labour market changes from immigration policy changing so much in the last few years. Those difficulties notwithstanding, the BLS has repeatedly failed to accurately estimate labor market conditions in the wake of the pandemic. Earlier this year, estimated job growth for the 12-month period ending in March 2025 was revised down by almost 900,000.

This helps explain why people were so sour on the labor market in polling data leading up to the 2024 presidential election, while the official statistics were painting a fairly rosy picture. After including not only monthly but also annual revisions to the data, it turns out that half of the job growth during Joe Biden’s last year in office never existed.

Instead of nonfarm payrolls rising by almost 2.4 million, they increased by about 1.2 million. Astonishingly, 48% of job growth was revised away, and the situation has not improved. The number of jobs in the economy at the end of 2025 has now been revised down by almost 1.1 million—meaning almost no jobs were added to the economy last year.

In short, the statistical problems that first reared their heads in the spring of 2022 still haven’t been addressed. That’s unacceptable. What’s particularly egregious is that so many folks outside of the BLS have been pointing out these problems for nearly four years now—and proposing solutions—and yet the problems aren’t fixed.

Even the Federal Reserve recently noted that they couldn’t trust the government’s monthly job estimates. That’s particularly troubling since they’re trying to set interest rates and conduct monetary policy, which has far-reaching implications for everyone in the economy. The central bank is flying blind because these data are untrustworthy.

By the time the BLS corrects the error, it’s too late, especially when monetary policy is concerned. That’s because it takes almost two years for the labor market to feel the full impact of a decision by the Fed today to adjust interest rates or engage in sales or purchases of financial securities.

And it’s not just the Fed that has been pointing out the problems with our government statistics. Countless investment houses on Wall Street have been sounding the alarm for almost four years too, in part because they have a vested interest in sound data. We cannot make sound business decisions if we have no idea what’s really going on in the economy broadly and the labor market specifically.

Part of the problem stems from the archaic way in which the government gathers much of its economic data today, relying on surveys largely like what was used a century ago. When more comprehensive data, such as quarterly unemployment insurance tax information, is finally aggregated long after the fact, the figures need to be revised.

In an age of real-time data, it’s inexcusable to be surveying a tiny fraction of total employers every month. Payroll processing firms like ADP and Paychex have a treasure trove of labor market data that the government is ignoring instead of incorporating. The data is already out there; it’s just not being used.

But getting more comprehensive data is only the first step. The BLS must also completely rework how it processes data too. For example, by wildly overestimating how many jobs are created by new businesses, BLS has inflated the overall number of jobs, only to revise away that job growth at a future date. Having inaccurate data means policymakers and firms are flying blind and have virtually no chance of making optimal policy and business decisions. The misallocations of resources that follow ultimately result in fewer jobs and less wage growth in the long run, hurting everyone. We can’t truly correct the economy unless we correct the data.