Skip to main content

Spend Analytics in the Federal Acquisition Sector

Acquisition in the federal government has been identified as a major source of potential savings in the future, as in the private sector, but a recent set of workshops and interviews identified some major challenges that exist.  First, the majority of data exists in a database called FPDS, or the Federal Procurement Data System, which was created in 1979.  This was a system written in COBOL, and was largely used for transactional purposes.  In 2006, the Federal Funding Accountability and Transparency Act called for searchable website database, leading to FedSpending.org.  In 2008, the Office of Management and Budgets created the database USAspending.gov.  Although intended to be able to provide transparency into federal spending, USAspending.gov was widely known to have major flaws and gaps.  For instance, the website claimed 0 dollars spent on Medicare in 2011 and 2012.

There are multiple challenges with FPDS data.  First, over 40% of FPDS data is not categorized.  GSA does not have control over much of the data that goes into FPDS and the codes used by agencies, unless this becomes a centralized function.  The product service codes are also outdated, and do not allow for correct transactional coding.  The system is also unable to split out codes within a transaction.  For example, a Marine Navy Internet technology transaction will have a single PSC codes, and cannot recognize spending on hardware, software and services separately.  Similarly, a construction PSC code cannot break out materials, labor, project management, and other services.

It is also difficult to get transactional data from suppliers.  Suppliers are generally not forthcoming with data, unless it is stipulated in contracts that data on spending be provided.  This is an opportunity for the future, as this requirement would need to be written into new contracts with suppliers, and this is currently not happening.  Future contracting practices should mandate specific formats and types of data that suppliers should be prepared to submit on a timely basis.  This is happening in a few remote sectors, but is not a standard practice.

Additional challenges for spend analysis of federal acquisition elements is driven by the fact that there is no easy way to reconcile and audit transactions.  Because invoices are in a different format and may not be tied directly to a specific PSC code, it is not easily amenable to a “three way match” that is traditionally used in a procure to pay system.  Many of the problems originate at the user level in agencies, where there is significant variance in the level of accuracy and precision when requisitions are entered into the system.  In addition, there is considerable variance in the level of maturity and ability of agencies to conduct spend analysis on their data.

The DATA Act

In May 2013, the Digital Accountability and Transparency Act was unveiled by the House Oversight and Government Reform Committee.   This act mandates that the federal acquisition system will provide information about budget authority, obligations, and outlays on the agency, agency component, appropriations account, program and object class level.  It is also intended to combine transaction-level obligation information (contracts signed, grants awarded, loans made) with outlays (actual checks cut).  Another important component is to assign universal unique identifiers to contracts and grant awards, establish government-wide data standards, and work to reduce improper payments.  There is also a provision to create a pilot program with contractors to examine the feasibility of recipient reporting of funds received, and it would transfer responsibility for running USAspending.gov from OMB to the Treasury.

Analysts in the federal government conducting spend analytics have had to rely on a variety of innovative data mining approaches to portray spending priorities and patterns, and are making headway in this direction as acquisition becomes more of a priority in the federal government.