Four issues with the USAF's tanker RFP, Part Three: alternatives for competitive OTS procurement

3.2.2.6.1    The KC-X shall provide a potable water system that supports maximum passenger load, patient, and aircrew requirements with running water available at each sink for 12 hours (473 milliliters per person per hour). (MANDATORY)

The boldfaced, block-capitalized emphasis in the extract above is the Air Force’s, and not mine. It is relieving to know that the service is minding the taxpayers’ dollars, and isn’t insisting on a rate of 474 milliliters per hour. Seriously, who is writing this stuff?

As strange as the current draft of the KC-X RFP really is, it is good to hear, as Aviation Week reported the other week, that the Pentagon is now backing away from its notion of an eighteen-year fixed price deal for new tanker aircraft. That approach constituted a 101 mistake in supply chain management, so it is good to see it gone. All the same, there would still be much work to be done if the government wanted to produce a sound RFP. As I will discuss below, it is not entirely clear that a sound RFP is the government’s objective. If it were, as I noted at the end of my last column on this subject (part two), the USAF could yet approach the problem with a radically simpler RFP, and one that induces much greater ex ante competition than the draft on the street today. In short, the Air Force should think about buying airliners a bit more like we all buy cars. I’ll get back to that at the end, but to show the logic, let’s work through the math, so to speak.

We can think about this in the familiar terms of supply and demand. In the classic market-clearing supply and demand model, sellers are generally willing to offer ever-increasing quantities only in return for higher prices per item, because eventually, all their production systems encounter diseconomies of scale.

This model isn’t so useful, however, in evaluating markets in which buyers need to choose whether to buy a lumpy quantity of a product or service, or not to buy it at all. Mass transit is a common case application: potential riders don’t select how many miles they want to travel to work, but whether or not to drive the car instead. The situation is actually similar to that in which the USAF finds itself today with the KC-X: the service wants to buy a fixed quantity of 179 Airbuses or 179 Boeings. In our model, we’ll consider a quasi-market with just one buyer and the two sellers, which we’ll transparently call A and B. As I’ll discuss further on, that’s rather close to the case in many aircraft procurement situations. Analyzing that discrete choice requires a hedonic model—one which matches price not to quantity, but quality. [1] Quality here is an aggregate of all the measurable qualities that a product might have, whether fuel capacity, cargo capacity, or fuel economy. It’s thus a rough measure, but a usefully illustrative one. In keeping with the USAF’s logical emphasis on the total cost of ownership (or in the service’s non-standard use, total adjusted price), we’ll also substitute TCO for price on the vertical axis.

The approach is certainly different; to begin, we’ll need multiple, firm-specific supply curves, because we’re trying to determine which firm will be doing the supplying. In this business in particular, the supply curves are essentially fixed in the short-term, as they are captive to the highly specific production systems of the aircraft manufacturers. They also show sharply increasing price-for-quality ratios, in that each production system can only customize its product—with more or less quality—to a certain degree without running up against the engineering limits of the design. An A330 or a 767 can only lift so much. With dissimilar designs, the supply curves probably don’t overlap: it’s essentially meaningless to wonder whether an A330 can haul as little fuel as a 767, and impractical to imagine, alternatively, how it could burn as little fuel per mile. There are, after all, differing aspects of quality endemic to both aircraft.

The demand curves are similarly specific to individual buyers; actually, we can draw multiple demand curves—indifference curves, actually—of increasing preference cascading into lower total costs for the same level of quality. In contrast to the supply curves, though, they are rather flexible in the short run, as they are driven by political objectives grounded by military judgement. That is, any given government or air arm, in formulating a corporate opinion about the sort of kit that it needs, comes to some decision of how much quality it prefers for how much money. The market clears where the last demand curve of increasing utility tangentially crosses one of the supply curves—thus identifying the winner of the contracting competition.

If the supply curves are fixed in the short run, then the customer’s decision—whether explicit or implicit—about the shape of his demand curve determines the winner and what will be bought. A slight shift in cost-sensitivity can hop the buyer’s preference from one supplier to another—even with a rather dissimilar product.

So much for the theory, which has been used widely in applications diverse as real estate appraisal and consumer price index adjustments. Its applicability in this case should be clear, but actual practice is rather curious. In the USAF’s KC-X RFP, rigid adherence to the fixed price approach has produced, whether the USAF’s strategists realize it or not, a nearly vertical demand curve. Leaving aside for the moment the analytical messiness of the one-percent-price-difference tie-breaker, the USAF is essentially saying that it has no demand for anything more than a fixed level of amalgamated quality, and will pay whatever it takes to get that (even though there must be some unstated budget limit). If, as is widely supposed, that level of quality is more closely matched to one of the pre-existing products being customized for the competition, it’s hard to imagine how the seller with the other product can successfully bid.

In the chart above, the price difference between what A and B can offer is pretty significant, and this is not mere modeling. With Northrop and EADS’s previous bid in hand, and an effectively vertical demand curve spelled out in the RFP, Boeing can raise its price and resulting TCO considerably, towards the lowest cost that B could offer under an RFP favoring A. Boeing lists its range of prices for 767-200ERs (the basis of the KC-767) as $127.5–139.0 million. Airbus lists its range of prices for A330-200s (the basis for the KC-45) as $176.3–185.5 million. We’re not certain as to the effective average price, or how they might differ, if Airbus and Boeing have differing discounting policies. Still, the effect is considerable. Excluding differences in how the offerors may choose to outfit their aircraft, the gap between the ends of those ranges is $37.3 million per plane, which gives Boeing a great deal of room for raising its profitability. After all, for all Boeing’s declarative strategy of launch-customer pricing, the actual pricing at launch may vary considerably with the likelihood that Northrop and EADS will actually submit a bid.

For the USAF’s attention, I will note that over a production run of 179 aircraft, that’s pricing headroom of $6.67 billion. One can call that the maximum effective price of writing a protest-proof RFP. It could be much less, as any offeror who wants the contract will be required to spell out its costs, and take no more than a 12.5 percent profit. However, in the absence of a bid from Northrop and EADS, we’ll never quite know how much less, as Boeing will have far more control over what is an allowable costs than the fact-checkers at the Defense Contract Audit Agency. It’s advertised costs may bake-in all sorts of minor inefficiencies to claim more dollars within that allowable 12.5 percent margin. And why not take a merely reasonable, rather than a vigorous, approach to cost-accounting? Boeing has shareholders, and the management they hire have a fiduciary responsibility to them. Driving a hard bargain is just good business. 

On the government’s side, doing good business means not getting taken by a supplier who knows that your BATNA—procurement shorthand for best alternative to a negotiated agreement—isn’t so good. The USAF could continue to fly airplanes that are already forty-something years old, and hope that none of them fall out of the sky in a flaming ball of JP-8, but that’s slowly becoming an increasingly tense position with time. (Bob Gates has signaled as much with his insistence that the competition get underway now.) Moreover, if Northrop and EADS do not lodge an effective bid, whatever extra profit Boeing accrues from the lack of competition will come without added quality in the product. The KC-767 is a fine plane, but there’s no reason to pay more for it.

So what to do? There are five steps that the government can take, on a short schedule, to improve the process, and almost certainly the outcome as well.

First, offer to pay for both bidders’ proposal costs, up to some prearranged and openly declared limit. Northrop and EADS’s disinterest in bidding is probably more than a gambit. Bidding on a tome of a requirement is a costly process. If there is really no real chance of winning, then spending the money on that such a bid would be economically unjustifiable. Even if Northrop and EADS are unlikely to win, inducing them to bid by offering to pay their reasonable bidding costs (or by just offering a flat fee for a reasonable bid) could inject a little more competition into the deal, and shave some basis points of  Boeing’s number. And if Northrop and EADS turn out to be stalking horses, then at least they won’t be able to complain after the fact about having been used: they will have been paid for his services.

Second, survey industry practice. Consider how this is done by people who buy airplanes for a living—the aircraft buyers at airlines. When airlines need new airliners, they generally do not release detailed RFPs to aircraft manufacturers. Some of these airlines have been bigger buyers of airliners for some time than the USAF, so their practices, melded by the discipline of the market, may be noteworthy. Airlines know full well that no manufacturer is going to design a plane from scratch for any single customer, so they recognize their choices are generally discrete, and often pairwise: A380 or B747, A330/340 or B777, CRJ-200 or ERJ-145, etc. In each market segment for commercial aircraft today, it is unusual to find more than two companies offering a product, and these products are readily identified. There is thus no reason to produce a detailed RFP, much less one specifying the flow rates of water in the heads, because these are what they are, and we can all read the specification sheets. RFPs are consequently broad, akin to asking we need roughly these kinds of airplanes, and about this many; what can you do for us?

Third, understand the economics. With this grasp of industry practice in mind, draw one more chart—a chart that conceptualizes how one would write the RFP. For broad guidance, the offerors will need the outlines of the qualities desired: minimal thresholds, maximal objectives, and the budget limit that they could infer anyway from the recently-published program objective memorandum (POM). What they should not have is the actual demand curve on which the government’s people will settle; that is, they should not have the actual grading criteria. With those criteria, the bidders could—as they can now—infer clearly which company’s product had the advantage, and the advantaged bidder could shade its bid accordingly.

Fourth, write a real RFP. To avoid this outcome, the government should consider writing an actual RFP. With so many mandatory criteria, and so little room to propose value-added improvements to the customer’s conception of what he needs, the request is not so much a request for proposal as a frighteningly complex request for quote (RFQ). In procurement theory and practice, the RFQ is much more useful when a customer needs an efficient arms-length transaction with a supplier for a relative commodity. Aerial tankers, however, are not resins. Specifying in so much detail both what is needed, and how much the customer is willing to pay relative to the next best option, surrenders a valuable informational asymmetry that could be used to extract competitive energy from at least one of the bidders.

Writing a real RFP, and not just a monstrous RFQ, can take military officers and procurement bureaucrats out of their comfort zones. Of course, this happened in the early 1990s when Al Gore decided to reinvent government and government procurement. That was a few years after he invented the Internet, of course—and while his involvement in the later project was probably more significant, the results were still pretty impressive. The Joint Direct Attack Munition (JDAM) program stands out as a poster child for how to do this right. The 1994 Federal Acquisition Streamlining Act specifically authorized the Secretary of Defense to waive any regulation not required by statute that might impede the efficiency of the contracting process. The JDAM System Program Office was then exempted from 25 provisions of the federal Acquisition Regulations (FARs) and another 25 of the Defense Department’s supplement to them (the DFARs). In their place, the government wrote an RFP that was little more than a two-page performance specification. [2]

Fifth, don’t hold an single-round auction; instead, negotiate. The results of the JDAM competition were impressive, but how they got that way is instructive. The USAF’s initial price estimate, based on parametric studies of past programs, was $68,000 per guidance kit (in 1994 terms). McDonnell Douglas, which was eventually to win the program, hoped that it could build them for $40,000 each. The USAF, however, had lent each the two companies that survived the first round of bidding several officials for its integrated product team (IPT). When McD-D proposed a way to get the price down to $28,000 per unit, the blue-suited people on the IPT hinted hard to their corporate colleagues that such a number wasn’t good enough. Lockheed, that is, was coming in with a bid of $18,000. McDonnell  eventually figured out a way to combine the functions of several parts into less expensive assemblies, and won with a final bid of just $14,000. [3]

At the time this was a shocking idea—though only in the government. In many industrial businesses, the concept was already then well-known, and by now, it’s old hat. Today, a considerable literature suggests that two-way trust and an efficient degree of communication are critical to making buyer-supplier collaboration work. [4] That may constitute an inconvenient truth, as at this point there may be little trust left in this three-way relationship. Further, the importance of forming these teams well should not be slighted. As Terry Little, the legendary initial manager of the JDAM program, told Government Executive magazine back in 1999, “a lot of people think being successful as a program manager is about making decisions, giving directions, good planning and having a good contract. I find none of that to be true. I find it is about people.” [5]

Yet still, the collaborate, multi-round approach may still give the willies to government procurement bureaucrats raised on the arms-length, single-round, sealed-bid auctions. After all, there is recent evidence from federal timber sales in the US that open auctions which recurrently feature the same bidders lead to tacit collusion, and less advantaged outcomes for the government. [6] In this case, though, the stakes in this single competition are so, the inherent rivalry in the observable conduct of the participating firms so obvious, that any kind of collusion is almost unthinkable.

Of course, that conclusion is a matter of who’s doing the colluding. In theory, the process could get disconnected, perhaps with cases of wine from either the Loire Valley or the Columbia Gorge. As a friend once from the aircraft-buying business put it to me recently, the party favors have been known to include “tickets to Riverdance (boring), a Christmas ornament (blah), a bottle of armagnac (always good)... and of course lots of dinners at Ruth Chris.” Ski vacations in the Cascades or the Alps shouldn’t be ruled out, either—presumably there are lots of “aircraft procurement symposia” (wink-wink, nudge-nudge) in both regions. Factors driving procurement decision-making may also very well include less untoward intangibles like egos, Buy-American sentiments, stick-it-to-Boeing sentiments, or previous employment by or connections to the manufacturers. 

Those concerns, though, shouldn’t disqualify the approach, as the airline industry has to deal with this stuff just the same. As Stephen Goldsmith, a former mayor of Indianapolis and current professor at Harvard’s Kennedy School of Government once put it, too much of government procurement regulation is still designed to keep civil servants from stealing money. [7] Darleen Druyuns are few and far between, and when they’re found, the regulations still are infrequently enough to keep their hands off the scales and out of the tills.

Finally, the concept of vague requirements, a hidden willingness-to-pay, and multiple rounds of negotiation shouldn’t be alien to anyone who has ever bought a car. Unless, that is, he bought all his cars from Saturn. That little division of GM that wouldn’t negotiate is no longer with us, and most everyone else has gone on haggling. There may just be some efficiency to this idea.

If the government really took this tack, would the KC-X program see a percentage of savings similar to that of the JDAM program? Of course not: the underlying products are far too mature for such a radical reengineering at this stage. Still, any effort expended on the heft of the revised RFP would be better spent in the analysis of the discrete alternatives available, and assembling government-supplier IPTs that could cooperatively refine the proposals into the best form possible, on each side of the bidding. There’s about six billion dollars on the table. This year, even in Washington, that’s real money.

NOTES

1. For the seminal paper on the concept, see Sherwin Rosen, “Hedonic Prices and Implicit Markets: Product Differentiation in Pure Competition,” Journal of Political Economy, January-February 1974.
2. A further two factors, rolling down-selection and contractor-supplied warranties, aren’t wholly applicable here. Excluding Ilyushin, are really only two suppliers capable to supplying the aircraft in question, so there’s no need for multiple down-selections. A warranty is an interesting idea, but since the maintenance costs of both 767s and A330s over time are readily knowable in advance, there’s little need, in terms of supposed informational asymmetries or quality incentives, to try to shift this risk from the customer to the supplier.
3. George Cahlink, “Birth of a Bomb: the Quest for Smart, Cheap, and Versatile Munitions,” Government Executive, 15 August 2003; and Cynthia Ingols and Lisa Brem, “Implementing Acquisition Reform: A Case Study on Joint Direct Attack Munitions,” Defense Systems Management College, July 1998. The JDAM procurement process was used for some time at the DSMC as a case study in how-to-do-things, so it’s not clear why the lessons have been forgotten so fast.
4. For a start, see Martin Hoegl and Stephan M. Wagner, “Buyer-Supplier Collaboration in Product Development,” Journal of Management, vol. 31, no. 4 (2005); Akbar Zaheer, Bill McEvily, and Vincenzo Perrone, “The Strategic Value of Buyer-Supplier Relationships,” Journal of Supply Chain Management, vol. 34, no. 3 (2006); and Chin-Chin Hsu, et al., “Information sharing, buyer-supplier relationships, and firm performance: a multi-region analysis,” International Journal of Physical Distribution & Logistics Management, vol. 38, no. 4 (2008).
5. Anne Laurent, “On Time, At Cost,” Government Executive, 1 September 1998. Terry Little is now a consultant with the Spectrum Group, so someone in the government ought to think about hiring him on this one.
6. Susan Athey and Jonathan Levin, “Information and Competition in U.S. Forest Service Timber Auctions,” Journal of Political Economy, April 2001.
7. Stephen Goldsmith, “Can Business Really Do Business with Government?”, Harvard Business Review, May-June 1997.

Comments

[Charles Armitage]

Thanks – I enjoyed that, although I found the theory a bit heavy going. An interesting parallel to “keeping the RFP vague” is the difference in the way the US and UK looks at tax avoidance. The US writes extremely precise and complex tax and if you can find a way round it, that is fine (until more precise and complex tax law is written to close the loophole). In the UK, the code is intentionally kept more vague, so the taxman can say “OK, I see where you are coming from, but no, I’m not going to allow it – nice try though”
 
The other thought is regards to the prices of the green 767 and A330 – the 767 is basically a dead programme (and has been so since 2004, when they produced a massive 8, or 0.7 per month or so) and is only being kept alive (at 1 per month) in the hope that they can win the KC-767, which looks as though it will run at 3 per month. The A330 is very much alive (currently at 8-9 per month) and is likely to continue at full rate production until at least 2015 (when the A350XWB is due to come in) and will continue for a while longer in freighter form – realistically I see the natural death of the A330 in about 2020.
 
This raises two points:

–         First, although the 767 line is likely to be pretty fully depreciated (EIS 1982), and although the A330 is probably getting there too (EIS 1994), you will get much higher benefits of scale from an aircraft line at 8-9 per month than 3 per month. In fact, I reckon that the fixed costs at Airbus (and Boeing) are about 30%, so as long as Airbus prices the green aircraft down by less than 30%, it will make more money than if they didn’t have the contract. This is not available to Boeing as the 767 line would be closed without the KC767. The difference in green aircraft price is thus probably not that much; on the other hand, the larger A330 will probably need more fuel, space etc., so the running costs per aircraft are likely to be higher. At the same time, the running costs per capability are likely to be lower.

–         Secondly, the USAF can either buy a brand new aircraft which should have been out of production for 5 years already (and is a 12 year older design), or one that commercially has another 10 years of life in production. Of course, part of the reason the 767 is on life support is that the A330 killed it (in the same way as the 777 killed the A340).
 
Regards,

Charles

[Alan Bangham]

Wow this is all pretty heavy analysis, and it is refreshing to see no bias on a quick review. Some thoughts...

Typically the KC-135 offloads 30% of its fuel payload, and the 767 carries a lot more than the 707 based KC-135, so why do we need a big giant replacement with the higher base costs and required infrastructure improvements? A 767 Tanker would have a cockpit more resembling the 787/747-8 than the current 767. A 767 based Tanker would be much more like the Navies brand new P-8 Orien submarine hunter, based on the 737, built on the same production line but not the same plane at all.

Big planes big wakes, just like a boat, simple rules of nature, physics. The U.S. military practices and conducts operations that no other country dares. For instance, night time carrier flight operations and below radar on the deck refueling at night. You want to do that behind a A330?

So if the USAF bought the A330, what are we gonna do, tear every single one down and then reassemble to make sure there are no Trojan Horses? This is a military asset.

And finally, we get nothing more than token help from Europe in Afganistan. They do not even vote reliably with us in the United Nations, so why would we be interested in exporting even one single job to Europe?

[Alan Bangham]

Oh yeah, how could I forget about the pending WTO ruling between EADS and Boeing, in favor of Boeing.

The nuts and bolts of that decade long dispute is that if EADS wants to build a plant anywhere in the US, all local and federal incentives are available to them. Not so for Boeing in Europe.