The Government Does Provide Data. It's Just Horribly Inaccurate.
So we're entering a new era in government transparency in which federal agencies put spending data online so the public can access, evaluate, and manipulate it, right?
That was the point of the Federal Funding Accountability and Transparency Act of 2006, which required that the government provide a "single searchable website, accessible to the public" that providing spending data for every Federal Agency.
Well, agencies are reporting spending numbers to USAspending.gov. The problem is they're not always on time. And sometimes the numbers are way off.
According to the Sunlight Foundation, it's a very big problem:
We've discovered that the data quality from the government is so poor that more than $1.3 trillion in federal reporting data from 2009 is wrong. These data inaccuracies account for 70 percent of the total $1.9 trillion in government spending data reported last year. Clearspending offers a critique on the reliability of data from USASpending.gov, across three metrics--consistency, completeness and timeliness--and covers spending from 2007, 2008 and 2009.
While there has been an increase in the number of programs reporting to USASpending.gov in the past three years, the reported data suffers from an abundance of errors, as well as problems with the data's timeliness and completeness.\n
You can check out the Clearspending scorecard here. The reporting does look appallingly messy—and it's not even consistent in its messiness. Nearly every department underreported in some areas, overreported in others, and reported late.
Clearly we have some ground to cover towards real transparency here. But it's important to recognize that it's only because the government is trying to provide consistent, complete, and accurate data that an organization like the Sunlight Foundation can point out that it's failing and hold it to a higher standard.
Here's my question: Are these agencies having such trouble providing good data because they don't have the right internal processes for reporting it or because they don't have good data in the first place? The latter seems like a much bigger problem.