Out of Scope, Out of Money

The U.S. Government Accountability Office (GAO) recently released a document detailing an investigation on the difficulties and costs associated with launching Healthcare.gov, the federal health insurance marketplace implemented as a result of the Patient Protection and Affordable Care Act (PPACA), informally known as Obamacare. The report, titled HEALTHCARE.GOV: Ineffective Planning and Oversight Practices Underscore the Need for Improved Contract Management provides a timeline of events that led to more than $200 mil (or almost 27%) in cost overruns, with the Government paying approximately $946 mil from fiscal year 2010 through March 2014.

GAO investigators focused this report on two task orders and an additional contract account for a total of $369 mil, which made up more than 40% of total contractual obligations reported by CMS as of March 2014. The two task orders required contractors to develop two core Healthcare.gov components – the Federally Facilitated Marketplace (or FFM) and a data hub, contracts awarded to CGI Federal and QSSI, respectively.

Federally Facilitated Marketplace System

The FFM system was designed to serve 3 main purposes: enable eligibility verification and plan enrollment, a plan management module for monitoring and renewing qualified health plans, and a financial management module that facilitate payments to insurance companies. With an initial bid of $56 mil back in 2011, costs for developing this system skyrocketed by the beginning of 2014 to over $200 mil. A short timeframe, unknown requirements, fluid legislation and contractor performance issues were noted as main contributors to this overrun, with authors noting that although CMS identified significant contractor performance, it took little action to remediate the situation, choosing to retain the same contractor in order to deliver on time (likely not only due to political and executive pressures, but conceivably also due to public opinion pressures).

The largest cost overrun occurred in September 2013, one month before the launch of the federal health insurance marketplace, to the tune of $76 mil, much of which was reportedly spent on bringing the system to a usable state in time for launch. After launch, CMS spent $1.5 mil to increasing storage capacity from 50TB to 400TB for the remainder of the development period. That’s almost $4,300/TB.

 Federal Data Services Hub

The data hub component acts as an information highway, routing and verifying information from the FFM against other federal and state data sources, like the Social Security Administration, Department of Homeland Security, Internal Revenue Service and more. This connection to other databases enables eligibility verification for both the purchase of health insurance, and the eligibility for federal subsidies. The initial cost estimate for the data hub was quoted at $30 mil, though latest reported figures show this to amount to more than $89 mil. While better planning, requirements gathering, and a better understanding of functionality lead to a lower difference between estimated cost vs. actual cost, this difference is still very much shocking and cringeworthy.

Again, the highest cost overrun was incurred just before launch of Healthcare.gov, with over $30 mil in additional costs being committed in September 2013.

A lack of communication, planning, accountability, and training

Report authors interviewed both CMS and contractor staff to better understand the complex relationships between the government and it’s contractors. Throughout the report, common themes begin to appear and unravel significant problems in the way government not only contracts with outside vendors, but also manages work and expectations. The lack of implementation of an acquisition strategy, a lack of clear organizational boundaries and responsibilities, a lack of training and communication, and improper/deleterious contractual terms led to the the government being cornered and obligated to pay for work not completed.

The work is not yet complete

While over 5 million Americans were able to enroll in a health plan through the federal marketplace, the work is not yet done, with many of the system’s core features still being developed. The FFM’s eligibility and enrollment service is only partially complete, with some elements not currently functioning and being completed in a new contract awarded to Accenture. Additionally, the financial management system, the least developed module, is not complete. While the American taxpayer has spent almost $1 bil to implement Healthcare.gov, much is still under development (read: currently being paid for). CGI Federal is no longer working on the FFM, and as of January 2014, a new one-year contract worth $91 mil was awarded to Accenture Federal Services to transition support of the FFM and continue development of what CGI Federal was unable to offer. Interestingly enough, this contract was also met with significant cost overruns, increasing to more than $175 mil after six modifications in a six month period. While CMS reported that this increase is only due to new requirements and features, GAO investigators mention that at least a portion of this cost is likely being spent on fixing previous issues.

Below are selected quotes from the GAO report that reveal the lack of expertise and experience that the Government has in contracting for software development:

Oversight Weaknesses and Lack of Adherence to Planning Requirements Compounded Acquisition Planning Challenges (p. 11)
Requirements for Developing the FFM System Were Not Well Defined When the Task Order Was Issued (p. 13)
CMS Did Not Fully Adhere to HHS Acquisition Planning Requirements and Missed Opportunities to Capture and Consider Risks Important to the Program’s Success (p. 17)
CMS Experienced Schedule Delays, Conducted Incomplete Governance Oversight Reviews, and Delayed Some Capabilities for the FFM and Data Hub (p. 23)
CMS Identified Significant Contractor Performance Issues for the FFM Task Order but Took Limited Action (p. 31)

Leave a Reply

Next ArticleCollecting Tweets Using R and the Twitter Search API