Treasury’s fiscal data site growing by leaps and bounds

The Treasury’s one-stop shop for data about federal finances has added six datasets since its launch last summer, with plans for a couple more additions in the next few months.

Fiscaldata.treasury.gov started with 18 datasets in July and is now up to 24 as the Bureau of Fiscal Service works to add all of its public-facing data to the site — potentially doubling the number of initial datasets.

The site already provides data-savvy users with comprehensive metadata and data dictionaries explaining the content, format, structure and use of datasets, and now officials want to go back and add more basic information — like explanations and context — for the general public.

“We really targeted data analysts for this initial site launch,” said Linnea Powell Xu, branch manager for data publication at the Bureau of Fiscal Service, during a Thursday webinar. “And now we’re going to be looking to expand our site to a broader audience and try and answer more of those general questions.”

The site’s Inside Fiscal Data section can help users answer questions like how much revenue the government collected in December ($346.1 billion) or how large the national debt is ($27 trillion). And BFS plans to add explanations of simple concepts like debt, revenue and spending.

But that doesn’t mean it won’t continue to answer new questions buried in the data, Powell Xu added.

The bureau maintains data on government-oriented categories such as Treasury securities, federal revenue and transactions, government payments, the public debt and also COVID-19 spending. It does not handle broader data about the U.S. economy or financial markets.

Users ‘know exactly what they’re getting’

Built by BFS from scratch with the open-source Gatsby code, the site came out of the governmentwide effort to increase public trust in agencies through easy-to-use platforms.

“What we wanted to do is really be a leader in moderating data practices as part of the Federal Data Strategy,” Powell Xu said. “So offering data in machine-readable formats so that users can easily get that data.”

The bureau used human-centered design when designing the site, including easily accessible application programming interfaces (APIs) and text search.

Datasets come from a variety of bureau systems and can be filtered by topics like “debt.” They can also be downloaded in CSV, JSON and XML formats, with detailed descriptions of each one provided.

“We heard in user testing that people often felt like they had to download the data to figure out what was in the data,” Powell Xu said. “So we wanted to make it clear this is what the debt-to-the-penny dataset is, and this is what it looks like, so users know exactly what they’re getting when they click that download button.”