Understanding the state and federal reports on the year-old American Recovery and Reinvestment Act is virtually impossible, despite the promise of "unprecedented levels of accountability and transparency" by President Barack Obama and Congress. Federal and state governments' efforts to inform the public often confuse, according to an analysis by the Maryland Public Policy Institute.

Not only are there discrepancies between federal and state information, but there are also major differences within federal reports on the official stimulus website, recovery.gov.

According to the interactive maps posted on the federal website in April, $3.7 billion was spent in Maryland from the stimulus bill by the end of 2009, creating 6,466 jobs in the fourth quarter. Maryland's stimulus website claims $4.1 billion was spent since the legislation passed, creating 19,967 jobs total.

Trying to get a single total is difficult because the government tracks both the amount awarded in a region (award amount) and the amount spent on projects in that region (local amount). In other words, local awards include transfers of stimulus dollars from one state to another, as when a project in a given state is undertaken by an out-of-state company.

Confusing matters even more, the mid-March version of the Maryland state summary on the federal stimulus website (available on the "downloads" page of recovery.gov) showed $8.6 billion total in local awards for Maryland by the end of 2009, creating about 300 more jobs than were claimed on the interactive map. When the federal government updated its spreadsheets at the end of the month, Maryland's total local award dropped to $3.7 billion.

What accounts for the difference between the state and federal numbers and the staggering revisions of federal numbers from month to month?

Cheryl Arvidson, assistant director of communications of recovery.gov, said there are a few reasons for the discrepancies, including some projects that probably should not have been reported in the first place. She also said the way the government calculates local awards changed during late March, and "we were still allowing regular updates of the previous reporting quarter data to fix mistakes that were either highlighted by agencies or by the recipients themselves."

But those issues can't explain losing $5 billion in local funding — more than the amount awarded to the state in the first place.

Many of the revisions to the federal government's Maryland state summary spreadsheet cleaned up mistakes created by companies in the initial report. For example, Sumter, S.C.-based Shaw Lumber Co. received $13,868 to deliver equipment to the National Park Service in Worcester County and reported creating one job. However, in its explanation for the type of job created, a company representative wrote, "I had to put a numerical in number of jobs to finish this report. We delivered materials only."

This mistake (and presumably others like it) has been removed from the updated state summary.

But even after the update, black holes exist throughout the data, making it almost impossible to figure out who received the money and for what purpose. And since the older spreadsheets are not available on recovery.gov, all that remains is the information reported by often-unreliable recipients. In the most recent Maryland state summary, only a little more than half the companies receiving stimulus money — 2,055 of the 3,623 recipients in Maryland — report any information aside from their names and the amount of money received.

"There is no easy explanation for why the numbers are changing, but they're certainly not making it easy for citizens to do data analysis," said Brian Riedl, research fellow in federal budget policy at the Heritage Foundation. "In defense of Washington, it is a rare thing to even be trying to do this. What they are attempting is hugely difficult."

Sandra Fabry, executive director of the Center for Fiscal Accountability in Washington, offers a different explanation. "It seems more a spin operation than a general transparency operation," she said. "It doesn't seem to be a genuine interest of theirs to have full data accuracy but to fudge the numbers so they can make their claim of saving jobs."

Whatever the reason, clarity is not easy to come by in the numbers. It is possible to glean tidbits of information about projects, but taxpayers may never find out how much they spent or how many jobs were created — or not — by the stimulus.

John J. Walters is a research associate at the Maryland Public Policy Institute. jwalters@mdpolicy.org.