Sunk costs

Sunk costs

Posted on December 21st, 2011 in Blog, News

Loren Thompson of the Lexington Institute has just published a piece lamenting the “$100 billion” wasted on weapons that were never deployed, or else deployed in much smaller numbers than predicted–thus not fully leveraging money spent on research and development. This is a huge issue, to be sure, but he frames it as resulting from politicians’ inability to consider “sunk costs.” If only that were true.

Thompson’s argument is that politicians live in the present (for the next election) and thus fail to consider the value and validity of choices made under past administrations. As a result, for “short term” budget considerations, they terminate or truncate programs that have not yet reached their potential, thus “wasting” the investment. Among the many examples he cites are Crusader, FCS, the Airborne Laser, the F-22, the Transformational Communications System, the Expeditionary Fighting Vehicle, DDG-1000, and multile missile defense programs.

Here’s the rub: taking into account the “sacrifices and assumptions of past administrations” (including, presumably, the money already spent) is exactly the wrong thing to do, for several reasons.

First, this turns the whole idea of “sunk costs” on its head. No business person who wanted to stay in business would justify further investment in a product line based on how much was already spent. Why should the government?

Second, the list of examples mixes apples and oranges. Cutting short a production run, while that increases unit costs, is not the same as ending years of Powerpoint developments or “science projects” that never produce an operational system. The reasons for these decisions are often very different.

Third, this ignores a very valid question about how much a system in development would cost to make it operational. It’s entirely possible that the benefit simply isn’t worth that cost. Alternative solutions may have arisen, technology may have changed or moved on, the environment may have changed, and so on.

Fourth, if anything DoD suffers from the opposite problem: an unwillingness or inability to kill off systems that are performing poorly in development. Sometimes this is for Service cultural reasons, sometimes for political ones, and sometimes it’s sheer bureaucratic inertia…but one could argue that a good chunk of the “waste” would have been avoided by pulling the plug earlier.

Most importantly, the problem itself is fundamentally different than the value (or lack) of past spending. Reasons that so much gets spent to so little result include a dysfunctional requirements system that cannot get the need right or resist changing specifications; overoptimism on initial costs and schedules that result in numerous extensions, overruns and missed milestones; resulting in an acquisition process that drags developments out well beyond original timelines. This creates a vicious circle: the longer a program gets stretched out, the less the need and environment resembles what it was initially. As a result, requirements change again, or the system gets “overspecified” to deal with more challenges than originally intended, increasing cost and complexity and further fueling the problem.

So I would argue that the real cause is NOT a system that undervalues sunk costs; on the contrary, it’s a system that perseveres too long and spends too much on poorly-defined and poorly performing programs. Rather than throw more money at these efforts to bring them to completion no matter what, we need a willingness to fail faster: to generate protypes and new ideas more quickly, try them out, expect more limited production runs, and move on. Both government and industry share responsibility here: government must fix the requirements process and provide incentives for these more rapid cycles. Industry must perform and deliver on what it bids. It’s a huge problem, but it’s not one that can be fixed by using sunk costs as the basis for decisions.