


It is the most anticipated global event every month. Thousands of analysts from around the world sit poised at their computers. Board meetings come to a halt. Government officials look at their phones. At 8:29 a.m., the world collectively sighs. It is “Jobs Day,” and the Bureau of Labor Statistics is about to release the monthly jobs report.
At 8:30 a.m. on the first Friday of the month, the BLS releases the Employment Situation report, a treasure trove of data about the U.S. labor market, including metrics such as total payroll employment, the unemployment rate, wage growth, and labor force participation.
Recommended Stories
- Project 2025 revisited: After 100 days, is Trump running the playbook?
- A perfect storm comes for elite universities
- Media's 'Maryland man' narrative collapses, but Democrats still coddle Abrego Garcia
The report’s headline numbers can send markets soaring or collapsing in a flash. Strong job growth and rising wages may signal inflationary pressures, prompting expectations of tighter monetary policy, such as interest rate hikes, which can dampen stock valuations and strengthen the dollar. Conversely, weaker-than-expected data may fuel hopes of rate cuts or stimulus, boosting equities but possibly weakening bonds.
The hundreds of data points in the report are then pored over by thousands of analysts who then make millions of trades worth trillions of dollars. The data in the jobs report can send oil skyrocketing and bonds falling, as well as crush companies and bankrupt nations.
But what if the data are wrong?
Not on purpose, mind you. It takes hundreds of people across two federal agencies to collect, analyze, and prepare the jobs report each month, and there is simply no way anyone could manage a conspiracy with that many people.
But most of the data underlying the jobs report come from two monthly surveys. A poll, basically, not too unlike the political polls Americans are inundated with every election season. And as anyone in the polling business will tell you, thanks to the disappearance of landlines and the unwillingness of more and more Americans to take a poll on their cellphones, the polling industry is in flux right now.
The old system of simply calling tens of thousands of phone numbers and hoping the first thousand who take the time to answer are somehow representative of the wider population simply isn’t working anymore. Just ask Ann Selzer, who, after decades of quality results, ended her career by completely blowing the final results of the 2024 presidential election in Iowa by more than 16 points. She had then-Vice President Kamala Harris winning 47%-44% when, in reality, President Donald Trump won 56%-43%.
What if BLS is missing the labor participation rate by 16 points every month? That could create some pretty wild volatility in world markets.
Fortunately, the Census Bureau, which collects the data for the current population survey of the jobs report, makes more of an effort to ensure its survey is representative of the population than Selzer did.
The Census Bureau’s current population survey comes from a pool of about 60,000 households recruited in person by census field representatives. Households are selected randomly by address, and then during the first visit, the demographic data are recorded.
After a household has been added to the pool, the census then contacts that household, usually by phone, for follow-up visits, once a month for four consecutive months. After that initial four months, households come out of the pool for four months, and then they become part of the pool for another four months, and then they are disqualified from participating. Households are added and subtracted from the pool on a rolling basis so that about an eighth of the pool is changing over every month.
If this sounds like a laborious process, it is, and its costs are rising. Travel and labor inflation have simply been rising faster than budgets. Then, there is the fall in response rates. As recently as 2013, about 90% of households in the pool completed the full survey each month. That number has fallen to 70%. That means there are now about 12,000 fewer completed surveys in the pool each month, thus weakening the reliability of the data. This is a particular problem for minority communities whose small size already makes their data vulnerable to large swings.
The census is now trying to improve its data collection efforts by moving to an internet self-response system that would function alongside the existing in-person and telephone collection systems. All of these systems are in desperate need of new software and hardware. And then there is the challenge of an aging workforce that is retiring rapidly before their expertise can be passed on to younger generations. Experts claim it will take about $25 million over five years to modernize this survey and about a $200 million investment to modernize all of the government’s leading surveys.
Another heavily anticipated monthly government data point, which has become more relevant is inflation has been eating away at consumer paychecks since President Joe Biden passed his COVID stimulus bill, is the Consumer Price Index. The Bureau of Labor Statistics manages and releases this data set as well.
There are two parts to CPI: First, the Consumer Expenditure Survey which involves interviewing about 24,000 households to find what goods and services (e.g. food, housing, transportation, apparel, medical care) households are buying and about how much of their budget each item is taking up. Second, the BLS collects prices for roughly 80,000 items and services monthly from about 22,500 retail and service establishments across 75 urban areas.
CPI is not the only measurement of inflation out there. There are other government and private measures that claim to more accurately measure inflation than the CPI.
The Federal Reserve for example uses something called the Personal Consumption Expenditure Price Index when making its decision about raising or lowering inflation targets. That data set is produced by the Bureau of Economic Analysis. But other differing methodologies aside, the PCE flat out steals about 25% of its data directly from the CPI. Without the CPI, there could be no PCE.
The same is true for private alternative measures like the Billion Prices Project and its successor PriceStats. Created by MIT economists, PriceStats scrapes prices from hundreds of online retailers covering millions of products to offer granular daily inflation data. But even PriceStats mirrors CPI’s selection of goods and services. Without CPI, PriceStats wouldn’t have the basket of goods to monitor or weight them to consumer spending patterns.
When it comes to inflation, no matter the data set, it turns out that it is just CPIs all the way down.
And like the jobs report, the CPI is also in need of modernization. On the survey/basket creation side, BLS is facing the same challenges as the Census Bureau is on response rates. To supplement the survey data, BLS is exploring ways of using credit card data to better reflect what people are actually spending money on.
On the price side, like PriceStats the BLS is developing its one web-scraping tool to capture price changes from online retailers. They are also exploring data-sharing agreements with brick-and-mortar retailers that would give them detailed sales data, reducing reliance on manual price checks at thousands of retail locations nationwide.
But again, these modernization efforts take money, money that Congress has proved reluctant to allocate.
The White House Office of Management and Budget has designated 40 monthly and quarterly reports as key economic indicators that are protected from early release to prevent market manipulation. These reports include the Department of Agriculture’s crop production report (of Trading Places fame), the Commerce Department’s New Residential Construction report, the Energy Department’s Weekly Natural Gas Storage Report, and the Federal Reserve’s Consumer Credit report.
DID GAVIN NEWSOM JUST DO SOMETHING USEFUL?
Federal employees caught disclosing any of the data from any of these reports before the report’s scheduled release can be prosecuted under the Trade Secrets Act and sent to prison for a year.
Each and every one of these reports has been deemed vital to both government policy makers and market actors. The desire to cut government waste and make our government more efficient is understandable and necessary. But the federal government does perform some core functions that are not best left to the market. Producing accurate data on the state of the economy is one of them. Congress and the Trump administration should be looking to invest in these functions, not cut them. Otherwise how will we ever be able to tell if any of the government’s policies are working?