Four issues with the USAF's tanker RFP, Part Two: an analysis of the split-buy concept

As I wrote in my last column on this subject, the US Air Force’s current request for proposals (RFP) in the KC-X tanker replacement program is troubling in four ways:

• It departs from value-for-money sensibility by equally weighting all the evaluation criteria
• It attempts to shift too much commodity and currency price risk to the supplier with a very long, fixed-price contract
• It pre-commits to a single supplier for that excessive duration
• It focuses on a perception of fairness to the bidders over value to the buyer

All of these problems could be dispatched by simply rewriting the RFP to reflect sound procurement practice. However, the Pentagon’s defense of the document has been as steadfast as it has been unsatisfactory, which is not generally a recipe for change. [1] It is entirely possible that the current RFP was hastily hatched from frustration with the GAO’s rejection of the Pentagon’s last attempt at grading papers. It has been suggested by supporters of Northrop and EADS’s bid that it is a veiled effort to steer the award to Boeing and its politically influential labor unions. Whatever the case, split procurement is not likely being proposed, even by its boosters, as the optimal solution, but the one with the best chance of successfully subverting the current process. Its efficacy and feasibility are further questions—and the subject of this essay.

The split procurement approach would effectively rerun a small competition every so often to keep, in theory, each contractor’s pencil sharp, awarding variable quantities based on each new bid and the relative past performance of the suppliers. It’s not uncommon practice in commercial procurement, though as I make clear below, there are good reasons that it’s not commonplace. Based on some of these, Ashton Carter, the US under secretary of defense for acquisition, and his boss Robert Gates have staked out adamant opposition to this idea. That said, split procurement has been recommended as offering at least seven advantages:

• Buying off political opposition
• Inducing potential bidders to bid
• Reducing bid prices (when managed carefully)
• Promoting a competitive industry
• Reducing the risk from correlated defects and uncertainty over future requirements
• Inducing faster innovation
• Delivering larger quantities faster

It is important to note that this is merely a relatively comprehensive list of the plausible arguments. To evaluate whether the approach is sound generally, and would further be useful in the KC-X competition, we will require some further discussion of the theory and evidence on hand.

THEORY

Buying off political opposition. Split procurement has been notably and recently forwarded by military and aviation analysts such as Loren Thompson and Richard Aboulafia as a convenient way to discourage further contract protests. [2] Using split procurement in this fashion is not, however, a new concept: the idea had similarly been suggested for the NATO AWACS program in the late 1970s. The British government (formed at the time by the Labour Party) insisted that the aircraft be based on the Hawker Siddeley Nimrod; the German and French governments had Airbuses in mind; but ultimately, the contract went to Boeing. [3] The story has obvious parallels to the last run of the KC-X competition beyond the mere similarity of the equipment: a rent-seeking company and its supportive labor union lobbied hard for a domestic solution, on all sorts of grounds supposedly “strategic”, but were ultimately rebuffed by a government which preferred value-for-money and good relations with trading partners and military allies. Further, while this move may be politically appealing, it is also, as I will explain further on, potentially politically dangerous.

Inducing potential bidders to bid. In part, this buying-off works by offering a consolation prize. When bidding is costly—and it is with large airliner projects—“the split award mechanism provides additional insurance against the possibility of losing the bid and, therefore, any bidding costs.” [4] Northrop and EADS seem at least interested in the concept, and generally speaking, contractors in two-horse races have often liked ongoing contract splits. As Bob Browning, CEO of Austal, said last year about the Littoral Combat Ship (LCS) program, “we'd rather have a 100 percent chance at building 27 ships than a 50 percent chance of building 55.” [5]

This may contravene a common, if simplistic, assumption in financial analysis of risk-neutrality. Ceteris paribus, the simple expected value of 27 certain units would seem slightly less than that of 55 half-likely units. In truth, though, few really expect the simple notion to hold, and for several possible reasons. Agency issues are probably present—managers like their jobs, many of which survive whether they build half the fleet or all of it. This builds in some incentive for bunting before swinging for the fences. Relatively concentrated ownership and control may play an issue as well, and in a similar way. Whether or not it is true about Austal or Boeing or EADS, there are certainly closely-held enterprises in which not all investors are equally widely diversified and risk-neutral. Investors who really are that passive aren’t likely to cause much trouble at the shareholders’ meeting, after all.

Expected values, though, are perhaps most notable for the Pentagon’s calculations. Offerers may think about these based on non-recurring investments. Assuming that the probability of half-the-deal were really 100 percent, those 27 ships in the LCS example could really be worth more than half as much as 55 awarded to a single company on a fixed-price, winner-take-all basis. That is because the Pentagon frequently pays contractors based on what they claim as their costs, and not what prices they can really deliver. So, if both contractors’ unit costs increase in a split award, the government could wind up paying for the same fixed costs twice—and for the margin that both earn on it.

Further, this stratagem only works when both contractors agree to play. After all, “a single bidder can effectively veto any non-sole-source spilt by submitting a relatively high price for its share of the split, thereby rendering the split unattractive to the buyer.” [6] That bidder might choose this strategy if he thought that his odds of winning all the production were so good that he need not offer to share it with his competitor.

Reducing bid prices (when managed carefully). Indeed, the usual approach to procurement assumes that buying from a single seller generally produces the lowest price. As alluded above, production scale economies suggest that the efficient outcome, ceteris paribus, will concentrate production in a single firm. This has been repeated argued theoretically in the literature, and shown empirically in military procurement in general, though with limited data sets. [7] That said, merely the threatened addition of a second source can be valuable to the buyer because it lends immediate credibility to the threat of switching suppliers. [8] Recent evidence from missile procurement in particular suggests that “dual sourcing as used in practice reduces procurement costs significantly, apparently spurring efforts at cost control that outweigh any lost economies of scale or learning that result from splitting production across two suppliers.” [9] Further, in a duopoly with long-lead contracts but soft current sales, one could theoretically do well to split procurement by capacity—each contractor might give a good deal to fill the remaining space in the factory, if other orders are unlikely to be forthcoming.

Promoting a competitive industry. Filling factories seems to have been a serious concern of the US Navy for some years now. Split awards have featured heavily in shipbuilding programs, but their track record has been mixed. Where required quantities are genuinely competed, as in the Arleigh Burke-class destroyer program, inflation-adjusted cost curves have often decreased over time—and shipyards focused on military production have remained in business for the next program opportunity. However, when the buyer has clearly signaled willingness “to pay a premium for a split award on strategic grounds, sole-source threats are not as effective in disciplining split prices.” [10] The Virginia-class submarine program may be the leading bad example here. Northrop Grumman’s Newport News Shipbuilding and General Dynamics’ Electric Boat Company were encouraged to divide the production of individual submarines between their yards in the interest of “preserving the industrial base.” If the next submarine program does not offer better scale economies than this one, then the current work sharing arrangement cannot be said to have eliminated competition in some hope of someday restoring it. Rather, it has likely induced collusion and increased costs to appease the politically connected.

What’s more, in the case of the KC-X competition, preserving competitiveness is at once unnecessary and unlikely. Both Airbus and Boeing reasonably can be expected to remain in the airliner business for some years to come, so neither will exit the market because of any purchasing decisions by this one customer. In the unlikely event that one actually did, it’s reasonable to expect entry into this segment, and probably from a company that already produces smaller airliners, such as Bombardier or Embraer. Further, the USAF’s projected quantities are actually quite limited compared to the necessary scale requirements of long-term, successful airliner production. [11] Thus, there is little room to expect that either contractor can or needs to be kept in the game with a hand-out award to promote some notion of a competitive industrial base.

Reducing the risk from correlated defects and uncertainty over future requirements. Second sources can be strong hedges against relying too strongly on any single supplier for quality and schedule performance. This is managed by gathering information over time related to correlated defects, and grading the contractors accordingly in the performance reviews. [12] In the KC-X competition, it could be particularly useful given Boeing’s recent past performance with tankers, which has not been fabulous. The company really did make a hash of its 767 tanker development for the Japanese and Italian Air Forces. Further, the problem was specifically with the refueling equipment, which is the core of what makes a tanker a tanker, and not just an airliner in military livery. [13] The same cannot be said about Airbus's recent work on its already-delivered A310 tankers for the Luftwaffe and the Canadian Air Force, or its A330 tankers for the Royal Air Force, the Royal Australian Air Force, the Royal Saudi Air Force, the UAE Air Force, and the Indian Air Force. While we should anticipate that Boeing will probably fix the Japanese and Italian problems convincingly, one cannot plausibly claim after these delays that the A330 represents a riskier option. But whatever the case, if either supplier’s aircraft showed systemic problems at some future date, the USAF would have more than one fleet of tanker aircraft (as it does now with the KC-135s and KC-10s) on which to rely, and running production lines from which to order more of the non-offending planes.

Material defects or delivery delays may not, however, be the most important risks to mitigate with split procurement. The strategic question of which aircraft is most suitable to US military needs may not be answered before quite a few years of production have passed. There are two aspects to this:

• Uncertainty over the need for fuel capacity. While dissatisfaction, if there has been any, over the KC-135’s fuel capacity did not drive the replacement decision in the first place, it is worth considering when re-equipping with a totally new type of airplane, as no one makes KC-135s (or 707s) anymore. The old Boeing’s fuel capacity was just about right for transpolar nuclear strike missions: each KC-135 Stratotanker would pair up with a single B-52 Stratofortress, and transfer as much fuel as it could through that high flow-rate boom before sending the bomber on. While there are other advantages, that is the leading reason that the USAF went with booms, rather than probe-and-drogues, which are cheaper and can refuel more aircraft at a time. Today, however, one can foresee potential advantage in the higher fuel capacity of the A330 (the KC-45). The forthcoming MQ-X program could be huge, and those unmanned reconnaissance-strike aircraft will frequently fly as long as they have fuel. The tankers will be in the fleet for decades, and we should anticipate that someone will figure out in the next few years how to get a drone to refuel reliably in mid-air. A relative shift in force structure towards drones could thus favor the A330; on the other hand, renewed enthusiasm for long-range manned bombers could, as outlined above, favor the 767.
• Uncertainty over the need for cargo capacity. Airbus has specifically been pitching the its tanker design as a Multi-Role Tanker-Transport (MRTT). The airplane carries all its fuel in the wings, as Airbus chose to put the larger wings of the four-engined A340 on the twin-engined A330-MRTT. That frees up a lot of internal cargo capacity, which allows for easy reconfiguration to lots of roles. Since the Defense Department appears intent on avoiding buying more C-17 Globemaster IIIs, whatever the intent of the Congress, there may be advantage in buying more cargo capacity. Most of the time, Air Mobility Command is flying stuff into long runways anyway.

Thus, the USAF could shade its grading in future competitions based of whether it found that it needed more or less fuel capacity, and more or less cargo capacity. But a reasonable counter-argument holds that the service could simply wait to discover whatever problems may emerge, and if necessary, switch suppliers at that point. Could not the USAF just be expected to do the right thing, and buy the right aircraft after its future requirements are further revealed? Cannot flexibility just flow from future intestinal fortitude?

Herein we might profitably be led, as the Viscount St. Albans wrote, to leave our managerial advice “much beholden to Machiavelli and others, that write what men do, and not what they ought to do.” [14] Establishing a new program in the presence of entrenched interest has often proven very difficult, so splitting the program ex ante could greatly aid shifting quantities ex post. Pre-committing to a diverse fleet now, to allow for a diverse fleet in the future, could rather be like the Pentagon’s privatization of much of its on-base housing in the 1990s. The government could have recapitalized its decrepit properties on its own, but the responsible offices in the Pentagon could never quite get the needed funding from the committees of the Congress that control military construction accounts. Privatization worked like a bond issue, mandating a certain level of expenditure over time for a guaranteed result now.

Inducing faster innovation. Pencils can be kept sharp not just for shading costs, but for designing the next new thing. Keeping more than one contractor in the game, the argument goes, can create a sort of internal market in which the competing suppliers repeatedly try to best one another’s innovations. Moreover, “when this incentive translates into large innovation expenditures and reductions in joint costs, the resulting price in a [split award] auction can be less than the price in the corresponding [winner-take-all] auction.” [15] Multiple rounds would thus keep multiple players in the game, and eager for business. On the other hand, second sourcing could reduce the innovative investments by reducing the suppliers’ rents from production. This has been long argued, dating back to Schumpeter’s claim that monopoly was the cost of progress, but the evidence is far from clear. [16]

Regardless of the theoretical efficacy, this argument falls a bit short when applied to the KC-X competition. Airbus and Boeing will continue to innovate, but the focus of those efforts will be on advances which will appeal to their more important customers, the airlines. Most of the value of those innovations will spill over into their military transport efforts, as air forces appreciate lower maintenance burdens, lower fuel consumption, and higher carrying capacity just the same. If a tanker fleet is eyed in the future for conversion to “multi-role tanker-transport-ISR” aircraft, Airbus and Boeing may offer innovative solutions, but so could Northrop Grumman or Thales. Installation of modular electronic systems on aircraft as large as these is a relatively separable matter.

Delivering greater quantities faster. In principle, engaging multiple factories will, as General Nathan Bedford Forest might have put it, build the most first. This was entirely the approach in the recent Mine-Resistant Ambush Protected (MRAP) vehicle program, in which the US Army and Marine Corps bought armored trucks in quantity for the same purpose from (depending on how one counts) three to five different contractors. Because the newest of the KC-135s are about 45 years old, there exists a prima facie case for replacing the aircraft relatively quickly—before one falls from the sky in a ball of fire. Dual sourcing, it is argued, could procure the planes twice as fast, if large quantities were ordered from both Boeing and Northrop. These could be variably awarded in the split procurement approach discussed, or they could even be awarded in even quantities, just to keep all the rent-seeking political, corporate, and labor groups happy.

The problem with this argument, though, is that dual sourcing may be useful for buying off political opposition, but it isn’t at all necessary to doubling the procurement rate. Boeing’s production of 767s has average 35 aircraft annually since 1982. As the RFP calls for about 10 aircraft annually, offering Boeing the opportunity to double production of a product that is otherwise being phased out would probably be met enthusiastically, and perhaps even with a decrease in unit price. Offering EADS the chance to build 20 militarized A330s annually in Alabama, instead of merely 10, would similarly draw no complaints of a lack of capacity. Boeing delivered 481 airliners last year; Airbus delivered 437. Neither company needs the other to relieve some supposed bottleneck that might result from another 10 orders.

EVIDENCE

That’s the theory, which one must say is mixed in its implications. So far, Ashton Carter and Robert Gates seem to be emphasizing the added cost and complexity of adding two types of aircraft to the USAF’s fleet, which can be easily imagined. Running a fleet with two new tankers (A330 and 767) and two old tankers (KC-135 and KC-10), each with parallel training organizations, parallel logistics flows, and different pools of aircrews and groundcrews, could seem daunting. This surely is problematic—but plenty of scheduled airlines fly several types of airliners. Southwest, as noted above, famously benefits from the simplicity of doing otherwise, but split procurements persist within organizations with considerably fewer aircraft and considerably greater cost pressures.

Just a few weeks ago, United Airlines announced its intention to buy billions of dollars of both 787s (from Boeing) and A350s (from Airbus) between 2016 and 2019. United’s shareholders don’t seem completely put off by that idea—rather, with extensive rights to cancel deliveries if the finances don’t work out, United will keep both manufacturers eager for its business. (This differs notably from keeping them in business, which as noted above is wholly unnecessary.) United may not be getting the launch customer pricing that Boeing is now proposing to offer the USAF, but it’s a pretty reasonable approach. [17] It’s particularly reasonable given that the Dreamliner only first flew last week, and the A350 is nothing more today than a so-called paper airplane. It also indicates how the generally efficient solution is not always and everywhere the specifically appropriate one.

Thus, we might conclude that the split-award concept is not a priori flawed, and that applying it to KC-X program would hardly be the “nightmare” that Bob Gates called it a year ago. [18] It is hardly a nightmare that the Luftwaffe flies A310-MRTTs while the Armée de l’Air next door flies KC-135s. These are much smaller fleets, but the investments are worth the money for air arms with different ideas about what a tanker ought to be—as the USAF might, if it wanted further options for tailoring its fleet to particular missions. The real question is whether, in the KC-X competition particularly, the sometime positive aspects of split procurement might outweigh the disadvantages.

ANALYSIS

To consider that question, let’s review how the theory applies. At least three of our original seven propositions don’t hold up very well in the context of Airbus, Boeing, and commercially-derivative military tanker-transports:

• Buying off political opposition
• Inducing potential bidders to bid
• Reducing bid prices (when managed carefully)
• Promoting a competitive industry
• Reducing the risk from correlated defects and uncertainty over future requirements
• Inducing faster innovation
• Delivering larger quantities faster

Given the very plausible threat that Northrop Grumman and EADS might choose not to bid, inducing potential bidders to bid would be primarily useful for, eo ipso, reducing bid prices. The alternative would be a negotiated solution with Boeing, but a solution in which Boeing would have no incentive to offer the USAF that launch customer pricing. After all, Northrop and EADS apparently underbid a possibly complacent Boeing by $15 million per plane in the last competition, and with a considerably heavier aircraft at that. [19] So, we might combine these two lines under the first heading, and proceed with a shorter list of justifications:

• Buying off political opposition
• Inducing potential bidders to bid, simply to reduce the prices bid
• Reducing the risk from correlated defects and uncertainty over future requirements

So what of those profound defects in the RFP? Do these managerial levers actually address them? The first tackles several, if indirectly. Buying off political opposition could slide the warring camps into shorter term contracts without an abrupt loss of face by the Pentagon’s leadership. This would eliminate the problems of misallocation of commodity and currency price risk, excessive pre-committment to a single supplier, and the obsession over fairness. The silly equal weighting of all criteria would need to fade away if the USAF came round to professing uncertainty over future requirements, and thus acknowledged that it would have to continually revise actual and meaningful weightings over time.

That’s how it could work, but that’s not necessarily working better. To begin, capitulating to transparent demands for patronage would make for stunningly bad public policy. If commercial and political interests in Washington State and Alabama could collude and conspire in this way, where would the line be drawn? We might then expect more federal congressmen to state baldly, as Massimo Fanucci might, that they would not need a lot, but that there could be trouble with political authority if they didn’t get it. Tolerating a little political rent-seeking may be the price of first-past-the-post elections; tolerating a lot would make the hullabaloo over earmarking oh-so humdrum.

The third lever seems intriguing, and hardly underhanded, and in one sense, the argument has already been accepted. There will almost certainly be a KC-Y competition in a few years, to replace the larger KC-10s, which can carry much more fuel and cargo than the KC-135s. Boeing would certainly have an opportunity in this competition as well, presumably with a tanker version of its 777, and Airbus would likely bring the A330-MRTT back again. If more, bigger tankers were considered important at that point, the quantities procured in the KC-X program could be held steady, and those in the KC-Y program could be increased. Indeed, the USAF might have considered holding these competitions concurrently, were not the KC-10s much younger, and had the federal military budget more slack.

With all this in mind, we might assess the split-sourcing idea thusly:

1. Just fixing the RFP could be easier, if the Pentagon can manage to get out of its own way.
2. If not, moving towards split sourcing could provide an excuse to rewrite the terms of the competition with criteria more conducive to keeping both contractors interested, which could have, as we saw from Boeing’s last bid, multi-billion dollar implications.
3. If split procurement is indeed pursued, the political appeasement angle must be downplayed to guard against a recurrence of this decade-long procurement debacle. The government should think seriously before offering any guarantees of minimum quantities.
4. In the end, procurement could effectively be split anyway, as Airbus could reasonably be favored to win the KC-Y competition, particularly if it occurs sooner than later.

There may be better strategies for tanker replacement than split procurement, but there are clearly worse ones as well.

PRACTICE

So how might this work in practice? While the Defense Department has run split procurements before, the peculiarities of the airliner market suggest that the government’s regulations for running these contests may not be wholly appropriate here. In my third (and perhaps final) column on this question, I’ll address how the USAF might approach the problem with a radically simpler RFP, and one that sidesteps the illusive notion of fairness.

———

Notes

1. See US Senator Jeff Sessions’ comments in Andrea Shalal-Esa, “Mobile, Alabama officials frustrated by tanker process,” Reuters, 5 January 2010.
2. Julie Johnsson, “Split could seal Air Force aerial-refueling tanker contract,” Chicago Tribune, 19 July 2009.
3. Arnold Lee Tessmer, Politics of compromise: NATO and AWACS‎, National Defense University Press, 1988.
4. Dorothy E. Klotz and Kalyan Chatterjee, "Variable Split Awards in a Single-Stage Procurement Model," Group Decision and Negotiation, vol. 4 (1995), pp. 295–310
5. George Talbot, “Murtha talks of splitting LCS contract,” (Mobile) Press-Register, 30 January 2009.
6. James J. Anton and Dennis A. Yao, "Split Awards, Procurement, and Innovation," The RAND Journal of Economics, Vol. 20, no. 4 (Winter 1989), pp. 538–552. In a subsequent paper, the same authors term this implicit coordination. See "Coordination in Split Award Auctions," The Quarterly Journal of Economics, Vol. 107, Issue 2 (May 1992), pp. 681–707.
7. W. R. Greer and S. S. Liao, "An Analysis of Risk and Return in the Defense Market: Its Impact on Weapon System Competition," Management Science, vol. 32, no. 10 (1986), pp. 1259-1273.
8. James J. Anton and Dennis A. Yao, "Second Sourcing and the Experience Curve: Price Competition in Defense Procurement," The RAND Journal of Economics, vol. 18, no. 1 (Spring 1987), pp. 57–76. For Anton’s most recent work on the subject, see James J. Anton, Sandro Brusco, and Giuseppe Lopomo, “Split-Award Procurement Auctions with Uncertain Scale Economies: Theory and Data,” Games and Economic Behavior, forthcoming.
9. Thomas P. Lyon, “Does Dual Sourcing Lower Procurement Costs?” Journal of Industrial Economics, vol. LIV, no. 2 (June 2006), pp. 223–252. For more on the subject, see William B. Burnett and W.E. Kovacic, “Reform of United States Weapons Acquisition Policy: Competition, Teaming Agreements, and Dual-Sourcing,” Yale Journal on Regulation, vol. 6, no. 2 (July 1989), pp. 249-317; and E. Pyatt, “Procurement Competition at Work: The Navy’s Experience,’ Yale Journal on Regulation, vol. 6, no. 2 (July 1989), pp. 319-331.
10. Sudhindra Seshadri, Kalyan Chattterjee, and Gary L. Lilien, “Multiple Source Procurement Auctions,” Marketing Science, vol. 10, no. 3 (Summer 1991), pp. 246–263.
11. The USAF has been down this path before, awarding contracts to McDonnell Douglas and Lockheed in futile attempts to keep them in the airliner business. See James Hasik, “Unstated and Unsuccessful: Observations on thirty years of the United States Air Force’s efforts at industrial policy in the transport aircraft industry,” April 2007.
12. Andrew Yim, “Failure Risk and Quality Cost in the Choice of Single Versus Multiple Sourcing,” Department of Accountancy, Tilburg University, 9 September 2009, p. 8.
13. Tom Kington, “Italian Air Force Nicknames 1st Tanker 'I Have a Dream',” Defense News, 17 June 2009.
14. Francis Bacon, The Advancement of Learning, Book II, xxi, 9 (1605).
15. Anton and Yao, op. cit., 1989.
16. For more recent arguments in favor of keeping contractors’ profits up to encourage future innovation, see Jean-Jacques Laffont and Jean Tirole, A Theory of Incentives in Procurement and Regulation, The MIT Press, 1993; Michael H. Riordan and David E. M. Sappington, “Second Sourcing,” RAND Journal of Economics, vol. 20, no. 1 (Spring 1989), pp. 41-58; and William P. Rogerson, “Profit Regulation of Defense Contractors and Prizes for Innovation,” Journal of Political Economy, vol. 97, no. 6 (December 1989), pp. 1284-1305. For an interesting view of how the trusts of Schumpeter’s time may have had lasting, beneficial effects, see Tom Nicholas, “Why Schumpeter was Right: Innovation, Market Power, and Creative Destruction in 1920s America,” Journal of Economic History, vol. 63, no. 4 (December 2003), pp. 1023-1058.
17. Amy Butler, “Split-Buy USAF Tanker Concept Gaining Favor,” Aviation Week & Space Technology, 29 January 2009.
18. “Albaugh as BCA head will be key to Boeing tanker bid,” Leeham News & Comment, 19 December 2009.
19. Leeham News & Comment, op. cit.