When administration officials this year sought data to learn more about the booming practice of interagency contracting, there was one place they couldn't go for the information: the government's own contracts database.

"We realized that the data may not be complete and accurate," said Robert Burton, acting administrator of federal procurement policy at the Office of Management and Budget.

Created in 1979, the Federal Procurement Data System (FPDS) was intended to provide the federal acquisition community with details on the government's $300 billion-a-year contracting operation. Policy makers could analyze the data and find out how a particular agency was spending its money or where it could be more efficient. Market researchers could develop insights into where the growth spots are for vendors interested in developing their federal business. And public interest groups could spot areas requiring greater oversight.

But shortcomings in FPDS are hindering policy makers and other users in their attempts to make procurement spending more efficient and responsible.

Policy makers and procurement experts grumble that the government's contracts database is unreliable when it comes to finding out almost anything about the government's procurement practices, such as how many contracts go to small and disadvantaged businesses, how much agencies use so-called performance-based contracts, or which Hurricane Katrina-related contracts were awarded without a competition and why.

The causes for the database's problems are many. People often enter contract data incorrectly because they don't understand the system. Much of the government's contract information gets logged into the database months after the contracting activity occurred. And much of the information that is logged in is erroneous or incomplete.

The government – and the contractor that manages the database – is trying to address these problems by automating contract data entry across government. In other cases, the government hasn't yet set up the system to capture data such as interagency contracting or the effectiveness of performance-based contracts.

In their search for data on interagency contracting, procurement leaders ultimately sent out lengthy questionnaires to each agency to get that information, delaying by months the formulation of new policies that would make the practice more accountable. Interagency contracts account for more than $32 billion worth of government business a year, but their use has been flagged by government auditors as a risky practice because agencies often misuse them.

Questionable data
There are other stories of how the database has not served its masters well. A 13-member acquisition expert panel – created by Congress to draft recommendations for how to improve the government's $300 billion-plus procurement operation – has tried using the database to research federal contracting.

In one case, the Acquisition Advisory Panel, as it's called, used the database to assess the government's use of performance-based contracts. Under performance-based contracts, agencies pay companies to meet specific service performance goals – such as answering customer calls satisfactorily within five minutes – rather than to perform specific functions, such as setting up a call center staffed by 30 customer-service representatives.

But the panel, after examining a sample of contracts labeled as performance-based in the database, found that 19 out of 48 contracts reviewed were not performance-based at all – agencies had wrongly categorized them in the database. And the database wasn't set up to answer key questions about performance-based contracting, a practice many contracting officials are struggling with.

"I don't know which [contracts] were failures and which ones were successful. I don't know which [contracts] met their cost standards or exceeded them," said panel member Carl DeMaio, president of the Performance Institute, a nonpartisan think tank, and chairman of the panel's subcommittee on performance-based contracting.

"FPDS is not a reliable database. It does not provide reliable data on federal acquisition and procurement. We've known this for many years," he said.

Frequent users of the database describe strange quirks in the data. Ray Bjorklund, senior vice president and chief knowledge officer at the market research firm Federal Sources Inc. in McLean, Va., said that he recently found the dollar value of NASA contracts declined by hundreds of millions of dollars from one year to the next. Several weeks later, after checking updated FPDS data, he found the amount had increased.

"If you don't constantly stay alert to the possibility that there may be big gaps in the data, then you may end up with the wrong answer," he said.

He said he didn't know what caused that swing in the data.

Another problem: A list of contracts for the Gulf Coast's hurricane recovery has scores of blank entries where it should say whether a contract was subject to competition, or the reason why it wasn't.

Many contracts are missing from the database because contracting officers are working without their automated systems, or have not had time to make entries, according to Global Computer Enterprises (GCE), the Reston, Va., company that has managed the database since 2003 under a contract from the General Services Administration.

How system works – or doesn't
Concerns about FPDS date back decades. The system began in 1978, when Congress set out to build a one-stop shop for data on government contracts. It was supposed to help the government set procurement policies by giving a timely, accurate view of spending.

GCE took over the system in late 2003 under the GSA contract and updated it to collect contract information as it is completed, rather than making agencies wait for the next periodic update.

Agencies use FPDS to see who they're doing business with, how much they're spending on various types of commodities or services, the level of competition in contracts, and whether they're meeting their goals for contracting with small companies. The system is meant to capture details on all executive branch transactions above $25,000, except for spending by classified programs and by a few exempt agencies such as the U.S. Postal Service.

The system has several dozen entries for each contract – the name of the contractor, the contract value, whether it was awarded competitively, whether it went to a small business, and other details.

At some agencies, contracting officials have to punch in the data; others have computer systems that automatically dump the data into FPDS as the contract is written.

Most civilian agencies now have automated systems, although they may not be used throughout the agency yet, OMB officials said. But the biggest spender of all – the Defense Department – has lagged for years in hooking up its contract systems into FPDS, the Government Accountability Office said in a September report.

That makes it hard to take overall snapshots of federal contracting. Bjorklund said it's tough to use the system to tell how much the government spends on information technology or other products, for instance.

Another result is that it's tough to peg the government's total procurement spending in a given year. Paul Murphy, president of Eagle Eye Publishers in Fairfax, Va., said he's had to bypass FPDS and go directly to the Defense Department for its spending numbers. Using those, he computed a governmentwide total of $377 billion in fiscal 2005 – a far cry from the $314 billion in overall government contracting in an FPDS report that shows how much of the money went to small businesses.

Bjorklund noted that the FPDS figure has likely grown as more 2005 records are added to the system.

Murphy noted that the data are getting better, but said the picture is still incomplete.

"Clearly, the trend is going in the right direction as far as completeness of reporting, but with [Defense] missing, you're still missing 70 percent of all the dollars," he said.

The information is faulty in other, smaller ways, such as with miscoded entries or empty entries, Bjorklund said. Both industry and government organizations use FPDS for many routine tasks, such as researching contractors, and determining which are the busiest contracting offices or what type of contract an agency uses most often, Bjorklund said.

But the people entering the data don't always know the nuances that can have a big impact on the final numbers. For instance, Bjorklund said, the data entry task may fall to an inexperienced user who doesn't know a laptop computer goes under computer services rather than hardware. Then the agency has a hard time telling how much it spent on hardware.

"Some of them are inexperienced, junior people, and haven't necessarily been trained on all of the intricacies of the data," he said. "If you don't really understand government procurement and the way government procurement can be characterized digitally . . . then you're going to be on a very steep learning curve if you're coming in for the first time."

Users say the system can sometimes give strange results.

"We think there are improvements to be made in the accuracy of it," said Fritz Trakowski, program officer for small business procurement at the Labor Department's Office of Small Business Programs. For instance, he said, the share of the department's contracts going to small disadvantaged business dipped from 4.7 percent to 1.6 percent from 2003 to 2004, before rebounding to 3 percent last year.

"I don't think we dropped three percentage points in one year. That's a huge change," he said. "I'm just suspicious."

His office is working to correct these suspected inaccuracies, he said.

Solutions
The Defense Department has been hooking up its users to FPDS in recent months. About 2,600 are linked now, 14,000 will be connected by summer's end, and Defense will start on the remaining 15,000 in October, with the goal of finishing them in three months, said Lisa Romney, senior procurement analyst in the Defense procurement and acquisition policy office.

Delays of recent years came from difficulties in reconciling Defense contract writing systems with FPDS, she said. Also, GSA needed to improve the system to make it easier to use and ensure that Defense and civilian agencies could access their data once it was entered, she said.

FPDS still will delay releasing Defense data for three months to avoid revealing information on military strategy.

In some instances, the system simply isn't set up to capture a certain type of information. OMB is working on adding interagency contracting information to the database so it won't have to go through such an arduous questionnaire process again, said OMB's Burton.

The agency issued a new manual for the system about a month ago that should clear up some of the confusion about how to record contracts, and will soon issue a change to the Federal Acquisition Regulation that clarifies who's responsible for entering data into FPDS, OMB officials said.

Burton said 24 agencies have finished verifying the completeness of their contracts data in FPDS for 2004 and 2005, and are now certifying it for 2006.

"We are optimistic that by the end of this year we can really say the FPDS system is vastly improved," he said.

DeMaio said agencies should keep closer watch over the data entry.

"FPDS data needs to have greater oversight, and sanctions if we have repeat instances of inaccurate data coming from an agency," he said.

He said he supports a congressional proposal by four senators – Tom Coburn, R-Okla.; John McCain, R-Ariz.; Thomas Carper, D-Del.; and Barack Obama, D-Ill. – to set up a public Web site that offers data on all federal grants and contracts.

Coburn, chairman of the Homeland Security and Governmental Affairs subcommittee on federal financial management, found it hard to use FPDS to get timely data needed to oversee federal contracting, said his spokesman, Aaron Cooper.

"It was extremely difficult and unruly a task to try to track down some of this information on government contracts," he said.

The bill, S 2590, was approved by the Homeland Security and Governmental Affairs Committee on July 27.

An official with GCE said the online tools for researching procurements have been revamped to be more user-friendly, and allow the results to be saved in another spreadsheet program, a change sought by many users.

The system is easily adaptable to new information the government says it wants to collect, said David Lucas, GCE director of business development.

"At some level, the system is only as good as the data that gets put into it," he said.

Regarding glitches in the data, he said, "We're always looking at the data for anomalies and to make sure that there isn't any widespread data integrity issue."

Lucas said the move toward automated data entry from the various agencies into the database is improving transparency in contracting.