I have been trying to understand what Southwest Airlines just did to its program and why. I believe that accounting rules, and to some extent reality itself, create sufficient incentives to destroy any "points as good as cash" revenue-based redemption program.

If miles and points were recorded at the time of issuance as a liability at their redemption value, airlines would have little incentive to "print" their currencies without limit. An overly generous frequent flyer program would be seen immediately as a loss-creating enterprise. But the accounting treatment does not work that way. Overly generous programs are common, with accounting rules concealing the losses for several years until reality creeps in and forces the airline to devalue its miles or points.

Let's look at the incentives created by the accounting rules. Most airlines account for miles or points in two buckets. The first bucket is points earned by flying. The second bucket is points sold to third parties, primarily banks.

Under IAS 18 Paragraph 19, points in the first bucket are NOT carried as a liability, except to the extent that an airline will incur small marginal costs to carry a passenger in an otherwise empty seat. You could look at this accounting treatment from one perspective as saying that a points redemption will NEVER displace a paying customer. Looking at it from another perspective, any displacement of paying customers is a non-event accounting-wise, since that customer's revenue was never booked in the first place.

In my opinion the first perspective is more accurate. A company which allows the use banked points to displace a large fraction of its revenue will go out of business. Foregone future revenue should not be ignored in present accounting, or else bad decisions will result. As they obviously have.

I should note that this incremental cost accounting for flight miles is used by most airlines, but that airlines are permitted to chose to defer more revenue according to the redemption value of the points, using the IFRIC 13 rules. You can guess why they do not make that choice.

Points in the second bucket are not carried as a liability, but neither is the revenue recognized until the points are redeemed or expire unused. This treatment, which is required by IFRIC 13, seems eminently fair, but there is one catch. If at any time the estimated cost of award redemption exceeds the money received for the points from the third party, the airline must immediately recognize a loss for the excess.

Southwest was the first large airline to adopt revenue-based redemption. I will use their program as an example.

We know that airlines sell their miles and points to third parties at around 1 cent each. How did it make sense in 2011 for Southwest to sell 1.67 cents of travel for 1 cent? Was Southwest merely borrowing from its future revenue at a high rate of interest? I believe they were.

Southwest presumably believed that people redeeming points would be taking trips for which they would not have paid the cash price. In a "points are as good as cash"revenue-based redemption program such as the current Rapid Rewards, that proposition is not credible. In the current (until April 17, 2015) program, each redemption reduces Southwest's revenue by 1.43 cents per point. (Southwest devalued its points by a ratio of 6/7 in March 2014.)

Southwest presumably also believed that few award tickets would displace paying passengers. When load factors were low, this belief was reasonable. With high load factors, a much larger percentage of redemption occurs on sold out flights, meaning that some revenue has been lost from other customers in addition to any lost revenue from the person redeeming points.

Loyalty programs face the challenge of providing customers with more than 1 cent of perceived value per point while losing less direct revenue than they gain by selling those points. In other words, loyalty programs need to create enough of a value wedge to be able to give the third party a cut of that wedge. It is not at clear to me that revenue-based redemption can ever be compatible with creating such a value wedge. To me, revenue-based redemption looks like a dead end, creating revenue today which is more than matched by decreased revenue in the future.

Southwest appears to have reached a similar conclusion. It has decided to tilt its redemption rates in favor of flights which will not sell out. This is a soft version of capacity controls, the lack of which was the prime selling point of revenue-based redemption. Current accounting rules seem incompatible with any program which does not include some form of capacity controls.

To the extent that redemption on 100% full flights decreases under Southwest's new rules, displacement of revenue purchases will decrease. However there is still the lost revenue from the member who is redeeming points. I don't see how one can deny that in most cases the member redeeming points would have paid cash for that flight. When you provide the member 1.43 cents per point of redemption value, you are losing approximately 1.43 cents of revenue from that member. If you only received 1 cent per point from Chase, you just lost 0.43 cents per point in the real world, regardless of how you do your accounting.

If you devalue points further and force the member to redeem at 1 cent per point, you eliminate this loss. However members will drop their affiliate credit cards in favor of cards offering 1% to 2% straight cash back. Chase will no longer pay 1 cent per point, and the whole scheme of selling points to third parties will collapse. It seems to me that revenue-based redemption puts the loyalty program in a coffin corner situation where it cannot devalue the points yet it cannot maximize the company's revenue either.

Chart-based redemption programs face some of these problems, but the charts offer members the possibility of outsized value for a particular redemption. In those cases the airline does not lose revenue because the member would not have paid the cash price and because the flight is not 100% full (as ensured by capacity controls). The classic mileage program structure is therefore sustainable as long as the awards attract sufficient numbers of people to affiliate credit cards.

One other detail. Southwest continues to offer zero-fee last minute redemption and redeposit. Paid tickets are similar, but the travel funds become limited to one year and locked to that one passenger, who might be an infrequent traveler. This difference make a points ticket slightly more valuable than a cash ticket. I estimate this effect at 5% or less in value, but it does contribute to Southwest's value wedge. Even if one believes that this difference will continue, I don't think it's enough to keep Southwest's program going in anything closely resembling its current form.

Is any "points as good as cash" loyalty program inevitably doomed? I look forward to readers' perspectives on this question.