From Hiro Aragaki:

First off, thanks to John Lande for pursuing this issue and calling attention to the real imprecision that sometimes attends our use of the term “BATNA.”  If anything, I have learned through my off-line exchange with him and Sanda Kaufman that there is more confusion out there among scholars and practitioners than I had originally anticipated.  Although I’m pleasantly surprised that John now sees things more or less the way that Sanda and I do, I think it is still useful to memorialize some key points that emerged in our off-line conversations.  I do this in the first three sections below.  This will likely be of most interest to students, practitioners, and teachers of the subject.

In the last section, I focus on what I think are some common misunderstandings about BATNA’s less popular (step?) sister, WATNA.  This will likely be of most interest to scholars writing in this area.  In the final analysis, I think John’s original complaint that we are using BATNA “wrong” may be better directed at WATNA.  I do think that many of us—myself included—have not been particularly clear about what we mean by WATNA, and in this sense may be using the term incorrectly.

1. BATNA vs. Expected Value (EV) or Reservation Price (RP).  This is a common mistake that my students often make:  They think of their BATNA as a number, such as $200,000.  John has found examples of scholars making this mistake as well.  But an ATNA (be it your BATNA or WATNA) is not a number or value; it is a course of action.  “Going to trial” or “purchasing the same trinket from another stall in the bazaar” are ATNAs.  “Two hundred dollars” is not an ATNA; it’s just a particular value one attaches to an ATNA.  That value, in turn, is helpful when trying to determine which is the Best or highest value ATNA.  To equate the EV of litigation with one’s BATNA or to equate one’s RP with one’s BATNA, however, is to make an apples-to-oranges comparison.
2. Valuing ATNAs based on “best case” v. realistic outcomes.  Take the example of a lawsuit in which the plaintiff’s best case recovery is $200k and worst case recovery is $0.  Through legal research, her lawyer has determined that there is a 20% chance that the best case will come true, a 60% chance that the worst case will come true, and a 20% chance of a third possibility: a $100k recovery.  The parties are now in settlement negotiations and the defendant’s last offer is $150k.  Based on these facts, my students will sometimes say, “My BATNA is $200k and my WATNA is $0.”

To begin with, this makes the mistake explained in section (1) above. An ATNA is not a number; it is the alternative itself (in this case, litigation).  Nor would my students’ mistake be cured by saying, “the value of my BATNA is $200k and the value of my WATNA is $0.”  The reason is that we are still talking about only one ATNA—continued litigation—so we haven’t yet gotten to the question of which is the Best from among the ATNAs.

In this example, negotiation theory would tell us to assign a value on the ATNA of litigation so that we can compare it against the defendant’s last offer of $150k.  Since we don’t yet know the outcome of litigation, we assign that value by looking at the most accurate or realistic prediction of the outcome (taking into account other things like risk aversion, transaction costs, other preferences for litigation vs settlement, etc).  We don’t do so by focusing exclusively on the best case outcome of $200k, which is what my students sometimes do when they claim that the value of their BATNA is $200k.  Why?  Because that’s just wishful thinking; it fails to take into account the very real possibility of recovering $0.  If $200k isn’t a realistic valuation of the ATNA of litigation, it’s hard to see how it is a valuation of a realistic ATNA at all, in which case it’s hard to see how it could represent the value of any particular ATNA such as the BATNA.  To be sure, negotiators will consider the best ($200k) and worst ($0) case outcomes when coming up with a realistic valuation of the ATNA of litigation, such as when they make the following EV calculation: $200k*20% + $100k*20% + $0*60% = $60k.  But that is not the same as saying they are considering their BATNA and WATNA.

3. “Best case outcome” within an ATNA vs. BATNA.  Even if we were to value the ATNA of litigation using best case outcome ($200k) or the worst case outcome ($0), it still would not tell us anything about whether that ATNA is a BATNA or a WATNA because we don’t yet know about the other ATNAs to which we are comparing it.  A plaintiff usually has at least two alternatives to settlement: continue with litigation or voluntarily dismiss.  Usually, continuing to litigate is better than an outright dismissal, which is why it is more or less accurate to call it the BATNA (or perhaps more precisely, the “Better” ATNA).  It may seem that the defendant has only one ATNA, but in reality there are many others one can think of: Giving the plaintiff everything in her complaint, secreting one’s assets, hiring thugs to kidnap or otherwise incapacitate her/her lawyers, etc. (some of which are not unrealistic in countries outside the developed world).  Here, too, litigation is likely the best of these alternatives, in which case it would be the BATNA.  In short, litigation is the BATNA to settlement not because we calculate its value by reference to the best case outcome but rather because we deem it to be the best from among the other alternatives to settlement.
4. The curious case of WATNA.  Virtually everything I said above with respect to BATNA applies, mutatis mutandis, to WATNA.  And if so, I now believe that WATNA is a surprisingly misused and useless concept.  This claim is likely to be met with some skepticism by many negotiation scholars. So I hope this post provokes some discussion.

Return to my example of the student who claims that her BATNA is $200k and that her WATNA is $0.  I argued above that neither claim makes sense.  John now agrees that the former is incorrect, and so I take it he would agree that the latter is also incorrect.  But here’s the thing: Many scholars and teachers would say that the WATNA is, in fact, $0.  One well-known source, for instance, urges us to “consider your WATNA (or the ‘worst alternative to the negotiation’)–what terrible thing might have happened if you left the negotiation and didn’t get what you want at all (such as losing a house, a deal, or the case in court.)”

But literally speaking, losing everything at trial is not the Worst from among other alternatives to settlement; it’s just the worst case outcome within the one ATNA of litigation.  Yes, negotiators will sometimes make decisions about whether or not to settle (or, for example, to purchase insurance) based on the worst possible outcome of unknown events.  But it is more accurate to think of that as risk analysis or management—which of course plays a role in valuing the ATNA—rather than as assessing the negotiation against one’s WATNA per se.

Assuming we are all rational utility maximizers, the very idea of a WATNA borders on the meaningless.  Consider this example: Seller is in negotiation with Buyer #1 and is deciding whether to accept Buyer’s last offer.  If Seller also has offers from three alternative buyers, we would expect her to decide whether or not to sell to Buyer #1 by comparing Buyer #1’s last offer against the Best from among the three alternative offers.  Would Seller ever have occasion to weigh Buyer #1’s last offer against the Worst of her alternative offers?  Not if she is a utility maximizer, which is perhaps “the” central normative assumption behind neoclassical economics and, by extension, much of negotiation theory.  And if all of this is accurate, I now find it hard to see how the concept of a WATNA adds any value to prescriptive negotiation theory at all.

Take another example: You are negotiating with the provider of long term care insurance.  You have received quotes from several other providers and have determined which of these is your BATNA.  Now consider what we typically think of as the WATNA—“what terrible thing might have happened if you left the negotiation and didn’t get what you want at all (such as losing a house, a deal, or the case in court.).”  In this example the terrible thing might be that you left the negotiation and before you were able to accept an alternative quote, you developed an ailment that requires long term care; as a result you have to pay for that care out of pocket.  Is that your WATNA?  I would argue “no.”  This eventual occurrence is not even an ATNA because, at the time you are negotiating, it is not a real or expected alternative course of action; it’s simply something that might happen to you sometime in the future.  In other words, it is a risk associated with terminating the negotiation and going to one of your other ATNAs—a risk that you should therefore factor into you RP.  But something that might happen to you in the future is not an alternative course of action from which you can choose.  So it’s not an ATNA.

The only actual ATNA that exists at the time you are negotiating and that is consistent with this terrible thing happening is the ATNA of doing nothing (i.e., leaving the negotiation and not accepting any of the other quotes).  But doing nothing is not necessarily the Worst of your ATNAs.  To determine whether it is, you’d have to compare (a) the likelihood you will succumb to a debilitating condition in the future multiplied by the costs of such a predicament without insurance, against (b) the total amount of premiums you would pay if you had accepted one of the other quotes minus the expected payout (I’m simplifying here).  So it’s not a foregone conclusion by any means that this “worst case scenario” if you were to do nothing is indeed the Worst from among your ATNAs.  You’d have to perform an EV calculation first.  We assume it is the WATNA at our peril and only because we are led to do what John earlier did with BATNA—that is, to equate the worst case outcome (in this case, of the ATNA of doing nothing) with the Worst from among all the other ATNAs (in this case, including the other quotes).

So, what my exchange with John taught me is that John was almost right: We are probably using one of these terms “wrong.”  But it’s not BATNA; it’s WATNA.

What say you all?