Why it matters that Trump is deleting government data
What happens when one of the world's largest repositories of free information becomes unreliable?
That dire situation came to pass in late January, when many federal websites took down entire datasets to avoid possibly violating a series of executive orders from President Donald Trump. At the same time, the Department of Government Efficiency, the Elon Musk-directed task force that is slashing government operations, has canceled contracts for the collection of data and cut workers who analyze and evaluate information about various facets of American life, such as education, housing and consumer protection. These developments led to widespread concern that some data that researchers have long relied on the government to collect would no longer be readily available, potentially hindering various initiatives meant to help Americans — both inside and outside of government.
Although some information has come back online, the availability and transparency of government-collected data is suddenly no longer a given. This isn't a worry just for data journalists like us, either. Experts I spoke with laid out how the public and private sector, companies large and small, and even the average person use government data to make decisions about program effectiveness, business prospects and life choices. With this in mind, I delved into this recent interruption in data access, why this data has significant value and what challenges may lie ahead if government data becomes less accessible and reliable moving forward.
Quantifying the data disruption
Since taking office on Jan. 20, Trump has signed a record number of executive orders, which tell executive branch agencies how to interpret existing laws when implementing policies. Among other things, these orders directed the federal government to recognize only two genders, pause foreign aid efforts and end diversity, equity and inclusion programs. To ensure they weren't violating those orders, many departments and offices took down datasets, fact sheets and other resources dealing with issues of gender and race. For instance, the Centers for Disease Control and Prevention took down data from the Youth Risk Behavior Surveillance System, which includes questions asking whether teens identify as transgender.
Across the government, agencies removed thousands of websites and datasets, from guidelines for vaccinations and avoiding sexually transmitted infections to data from the Census Bureau's American Community Survey. Some have since come back online; others have not. Concerned about the potential loss of critical information, many organizations and individuals around the country worked to preserve what might otherwise be lost by archiving the remaining datasets and recovering missing info from older versions of websites using tools like the Internet Archive's Wayback Machine.
One way to assess the scope of the losses is to look at the total number of datasets that are available on Data.gov, an official government website that catalogs thousands of datasets, predominantly from federal agencies but also other sources, such as local governments. On Jan. 19, the day before Trump took office, there were nearly 308,000 datasets on Data.gov, but that figure had fallen to around 305,500 by Jan. 22. In early February, that number increased again, although it has not returned to the pre-Trump total.
However, it's important to note that these topline numbers from Data.gov represent only a back-of-the-envelope measure of data loss. Some datasets linked on the site aren't necessarily available when you click on them, such as many hosted by the United States Agency for International Development, which the Trump administration has all but shuttered. Plus, the site's count has changed for many reasons over the years, such as the removal of older datasets, shifts in technology or changes to which non-federal organizations are linked on the website. For instance, during the early days of former President Joe Biden's tenure in January 2021, the number of datasets available on Data.gov remained largely static. But then the retooling of the site using an updated cataloging service led to a drop of around 25,000 in the raw count of downloadable files in early February of that year.
Government data helps power business
It's not just academics, journalists and government officials who were impacted by this loss. In our information-driven age, large portions of the U.S. economy use government data to make business decisions. "In their development of plans, businesses have to benchmark how they compare to the rest of the industry or in markets that they're targeting," said Ellen Hughes-Cromwick, a senior visiting fellow at Third Way. "You can't do that without government data."
Formerly the chief global economist at Ford Motor Co., Hughes-Cromwick knows firsthand the importance of government data to operating an automobile manufacturer. A paper she coauthored in 2019 presented a table of short-run and long-run indicators that a car company would rely on when making production decisions, which can involve data from at least nine different bureaus, departments or other entities within the federal government.
This data can help a carmaker determine its hiring, investment and production goals. For instance, when setting monthly production targets, a firm will review data on auto sales from the Bureau of Economic Analysis, data from the Federal Reserve on the production levels of all types of vehicles across the automotive industry and consumer prices from the Bureau of Labor Statistics. The company uses this data to set production levels at each plant, even the speed of the assembly line.
More broadly, Hughes-Cromwick stressed that leading indicators in the data can affect long-term business plans. Companies across sectors considering whether to expand a product line may look at unemployment insurance claims from the Department of Labor to see if the overall market is weakening (claims rise) or improving (claims fall). They will look at whether retail spending is up (good) or down (bad) using sales data from the Census Bureau. Businesses also look to datasets that come from nongovernmental sources but that are partly informed by government data, such as the Purchasing Managers Index, a measure compiled by the Institute for Supply Management that firms, suppliers and investors use to get a sense of current and future market conditions.
Sectors that are especially reliant on this data also affect huge parts of the economy and produce sizable output. Last year, the Department of Commerce estimated that industries that heavily rely on government data reached revenues of almost $800 billion in 2022, nearly twice as much as in 2012. In that time, government-data-intensive sectors grew faster than the economy as a whole, with a particular boost from internet-based publishing, broadcasting and search portals as well as management consulting services.
Government data is particularly useful to the financial services industry. Naturally, firms such as commercial banks, asset management companies and equity brokerages use this information to decide where to direct their investments in the economy. But they also use this information to stress test whether their current portfolios could withstand a major economic or financial shock to the system. Regulations requiring this sort of testing came about in the aftermath of the 2008 financial crisis to make sure systematically important financial institutions maintained sounder footing in the face of future adverse conditions.
Less data, worse public policy?
Agencies in the federal government do more with data than collect and collate it. They use that information to identify public policy problems, develop programs to address those problems and then evaluate whether or not the programs are working using an evidence-based approach built on rigorous assessment of the data. "It's this whole cycle that's really important, and that is, I think, in jeopardy right now," said Molly Irwin, former chief evaluation officer at the U.S. Department of Labor and a board member at the Data Foundation.
Irwin pointed to the labor market as an example of how this cycle works. The government collects jobs data, and the Department of Labor uses that information to identify gaps in the labor market where employer needs aren't being met. The department then funds grant programs to train workers to fill those holes. Once those programs are in place, the department evaluates their effectiveness using administrative data gathered by the programs. They also conduct studies, which can involve surveys and interviews with employers — in essence, collecting even more data that gauges whether a program is working. That evidence is used to determine how well a program is working and how it could be improved. A successful program could become an approach that others also use to fill future labor market needs.
However, Irwin told me the means for federal evaluators to examine program effectiveness are now up in the air due to funding and staff cuts made by the Trump administration. For instance, DOGE canceled around $900 million in contracts at the Department of Education's National Center for Education Statistics, spending aimed at collecting information on the U.S. education system — data that will now be unavailable for researchers hoping to improve educational outcomes. Staff cuts at the Department of Housing and Urban Development have eliminated positions that compile data on the housing market and study solutions to homelessness and affordable housing. The Department of Labor, meanwhile, has taken down databases that contain information regarding effective training and employment programs, including the Clearinghouse for Labor Evaluation and Research.
This seemingly runs against the goals of legislation that Trump signed into law in 2019, during his first term: the Foundations for Evidence-Based Policymaking Act. The legislation essentially laid out a system for collecting and accessing data that federal agencies can use to develop evidence-based policies. It compels agencies to submit annual plans for determining and addressing policy challenges to the Office of Management and Budget as well as Congress. In 2024, a survey of federal evaluation officials conducted by the Data Foundation and American Evaluation Association found that 83 percent reported that the Evidence Act had helped them achieve their missions.
"[The law] was really about building the capacity in the federal government to do research studies, to build and produce evidence and, importantly, to use that evidence to make evidence-based decisions," Irwin said. "Those are the tools that we are losing as the evidence capacity and the datasets and, in many cases, the reports of findings and all of that goes away."
The potential future costs
The prospect of government data becoming less accessible or lower-quality could have long-term consequences — both tangible, in terms of dollars and cents, and intangible, in terms of confidence in the data that helps run our world.
Government data is, practically speaking, free for businesses to use. Open data is readily accessible and paid for by taxpayer dollars. A company like Zillow, for instance, leveraged governmental datasets and market analysis to become a leading provider of real estate information for consumers — all while having a market capitalization of almost $19 billion. Acquiring alternative forms of economic and demographic data, if it's even available, could become a cost for companies that currently doesn't affect their bottom line.
Moreover, knowledge and confidence in economic data help drive investment choices. As Lizzie O'Leary at Slate recently observed, swings in the stock market and bond market affecting beaucoup dollars — not to mention the policies of the Federal Reserve — come in reaction to the monthly jobs reports from the Bureau of Labor Statistics. But with the interruption in access to some government data, the people who use that data — firms, brokers, analysts, economists — could become less certain about its availability and reliability, which in turn could affect their investment choices, which in turn could affect the wider market.
"For any economist, we need a time series to do statistical and regression analysis for a lot of microeconomic data and industries that we might be examining," said Hughes-Cromwick, speaking about the possibility of data from government agencies becoming less accessible. "They're running the risk of causing some discontinuous time series where we won't have measurements — let's say for March 2025 industrial production, or April 2025 employment statistics. That would be a significant breach in the quality and accuracy of our data."
And while cutting evaluation teams from government departments could save money at first, it could prove expensive in the long run. If government programs remain in place — and despite the Trump administration's intention to cut parts of the federal budget, many will survive — they won't necessarily be evaluated as rigorously moving forward. That might make it more challenging to improve the functioning of a program, or allow a costly but ineffective program to remain in place. Additionally, Irwin noted that the Trump administration's cuts have halted ongoing studies that in some cases were nearly complete. As a result, those findings may never see the light of day, losing any gain from that investment.
Now, this is not to say that improvements couldn't be made. "Certainly there are some more modern approaches to developing statistics that would help improve the quality and timeliness of the data," Hughes-Cromwick said. But she pointed out entities like the Census Bureau and the Bureau of Economic Analysis have worked to expand access to real-time data from the private sector as well as administrative data.
The broader danger to the reliability and availability of government data could make it even easier to distrust information in a societal environment in which conspiracy theories can catch like wildfire on social media and the internet. Faith in government data may already be flagging given that public trust in government is quite low: In 2024, only about 1 in 5 Americans told pollsters that they trust the government to do what is right always or most of the time, according to data collected by the Pew Research Center.
Lastly, the removal of government information could be felt by all Americans. For instance, states, localities and parents use information from the National Center for Education Statistics to understand the strengths and weaknesses of school systems. "The Nation's Report Card is something that parents use when they're thinking about, 'Where do I move?'" said Irwin. "'What are the statistics in different school districts so that I can move to a place to put my child into a school that is doing well?'"
"All of us use data and information to make all kinds of decisions every day," Irwin told me. "It's the same kind of information that we need at every level. How am I going to invest my personal budget?"
It's the same with the government, she added; it should have those choices to invest its much bigger budget that comes from taxpayer dollars. "We want the government to have that information so that they can invest dollars in the right things."
