Contracting for better GPRA results

Several years ago, the Department of Veterans Affairs contracted for information technology to help the agency reduce the processing time for veterans' benefits claims. At the end of the day, the vendor got paid, but processing times were reduced only minimally.

I relate this example neither to insult the VA nor to blame the vendor working on the contract. Rather, I bring it up because of its relevance to a wider class of cases.

Clearly, improving service to veterans was something of value that the VA was willing to pay for, even though the VA wasn't going to attain budgetary savings for itself as an organization if the processing time were reduced. This is a common situation. Benefits that IT and business process re-engineering (BPR) bring to agencies often are "off budget"— that is, they go to agency customers in the form of better service or higher quality. A second feature that the VA effort shared with many others in the government is that it failed to deliver the anticipated results, but the vendor got paid anyway.

Imagine if the VA had contracted for veterans benefits proc-essing time reduction in a different way. Imagine that the agency gave an average "as is" processing turnaround time in the request for proposals and explained how that time was calculated. Then imagine the VA invited vendors to bid a dollar amount that they would ask to be paid for every day they shaved off the baseline processing time. The more days shaved off, the greater the payment. No days shaved off, no payment. Think of the changes in incentives such a contracting arrangement would unleash!

Now let your imagination run wild. The methodology I just suggested could be extended to just about any quantitative goal an agency has set for itself under the Government Performance and Results Act (GPRA) where the agency feels that contractor support might significantly help it reach the goal. You want to think "out of the box"?

Take an agency that measures customer satisfaction and seeks to increase satisfaction as a GPRA goal. The agency puts its current customer-satisfaction measures (and measurement methodology) in an RFP and asks contractors how much they would expect to be paid for each percentage point increase in customer satisfaction that the contractors help the agency achieve— again, with no payment if no improvements occur. The same methodology could be used for an agency with a GPRA goal to decrease burden-hours placed on the public, and so forth.

This form of contracting would require some novel approaches to structuring the competition for the contract. The most important change would be to use the "due diligence" approach that I discussed in an earlier column [FCW, June 22] to give bidding vendors an opportunity to learn about the agency's relevant "as is" business processes.

Agencies also could choose many different approaches in inviting bids on payment schedules. My advice would be to give bidders the fullest possible freedom in offering payment schedules, which the agency would then evaluate— along with technical approaches and past performance— as part of vendor selection. Differing payment bids might include getting different amounts of money for different levels of performance improvement or allowing bidders to ask for some minimum fixed-price payment along with payments for performance improvements. The agency would need to specify the time period over which the payments would be made. The agency also might choose to specify a cap on the amount of money it would be willing to pay. If it did so, that would clearly affect vendors' bidding strategies; any such amount should be discussed with the vendor community during pre-solicitation market research so that there is not a radical disconnect between agency and vendor expectations.

This approach also will require restructuring the relationship between the government and the winning vendor. One could imagine this form of contracting being used for operations that are outsourced to vendors. But more frequently, the vendor will be brought in to apply IT and BPR to an operation performed by government employees. In such a situation, performance improvement will require efforts by the vendor and the government working together. This creates potential risks for both parties.

For example, the vendor risks not getting paid because the government flounders in implementing the vendor's IT or BPR solutions. Similarly, the government risks making payments to the vendor for improvements to which the government contributed and which were thus not brought about solely through the vendor's efforts. What is required for this arrangement to work is the kind of deep vendor/customer partnership that many have espoused in other contexts for bringing about the highest chances of success of any IT venture.

The ideas in this column are, at this moment, in the realm of my imagination. But I am convinced that linking contracting with GPRA goals in this way could be key to helping agencies improve their performance on many of their GPRA goals. Some day, if government is smart, this form of contracting will become commonplace. Here's my challenge to the many innovative people out there in the federal government: Be the first out of the starting block to help the government re-engineer its approach to contracting.

-- Kelman was the administrator of the Office of Federal Procurement Policy from 1993 to 1997. He is now Weatherhead Professor of Public Management at Harvard's Kennedy School of Government.