October 19, 2012

Why Scenario-Based Planning Beats Prediction

Patrick Marren
Partner

Scenario consultants have an advantage over single-point forecasters like Nate Silver: we're not restricted to single point forecasts. 

Don't get us wrong: Nate Silver is as good as it gets when it comes to single-point forecasts, and as honest about their limitations.  But that undeniable fact just further dramatizes how dangerous point forecasting can be.

Just to illustrate, let's take the current ongoing presidential hoo-hah. 

Silver's Five Thirty Eight blog, as of this morning, had President Obama's odds of winning the presidency at 70.4%. This is up from 61.1% a week ago. 

That's a big enough shift in and of itself. But it's also down from a high of 87.1% on October 4. 

So if you have high-stakes business decisions to make that depend on who gets elected president, what level of odds do you require to pull the trigger? 61.1%? 70.4%? 87.1%? 99%? 

This approach shows the limitations of the single-point forecast. Because if the stakes are truly high, even 99% may not be high enough. Even 99.9% may not be. One-in-a-thousand events occur every day, folks. (Every week or two or three someone wins Powerball, and that's a one-in-175 million chance.) 

When the level of risk (and uncertainty, which is different and a separate topic - see Silver's book The Signal and the Noise for an explanation), in combination with the potential impact, of a possible event is above the trivial, a single-point forecast is not a serious approach to decision-making.

What you need for the vast majority of serious decisions under conditions of future uncertainty is to abandon the single-point forecast approach in favor of a multiple-alternative-scenario approach. Because, let's face it: it is entirely possible for Mitt Romney to win the presidency, and it is entirely possible for Barack Obama to win re-election.

Or let's put it another way: Maybe Barack Obama is a lock to win, or maybe Mitt Romney is a lock to win, but there is no way for us to know beforehand which of these hypotheses is true.

Or let's put it a third way: Saying "The chance of Mitt Romney winning is 29.6% at this moment" is, in a sense, obviously wrong. Because the chance of Mitt Romney winning the presidency on November 6 is either 0% or 100%. He's not going to win 29.6% of the presidency. Nor is Obama going to win 70.4% of the presidency. It's all or nothing.

Any bet that is predicated on a partial win for one of them will be wrong. So what we really have here are two scenarios: MittWorld or BarackReElectedWorld. 

But let's take an even closer look. Is a Manichaean 2-dimensional look good enough to make decisions? So Obama gets re-elected; does that tell you enough that you are able to make serious decisions? I seriously doubt it.

Let's throw in another dimension: Sequestration/No Sequestration. There's a huge budget negotiation scheduled to take place after the election; if it fails, automatic huge cuts will fall on both entitlement programs and the military. What are the odds of the negotiations working out versus them falling apart and the mandatory cuts taking place? Again: They are either zero, or one. 

The same analysis applies here: a single point forecast is far too risky an approach to bet on. You need to plan for ALL FOUR possibilities: Obama/Sequestration; Obama/Budget Deal; Romney/Sequestration; Romney/Budget Deal. Each of these scenarios is plausible; each has different ramifications for many decision-makers in both the private and public sectors. Any serious person should be sketching out her/his response to each of them, because the real chance of each of those outcomes is either 100%, or it's zero, and we can not know until too late which it is. It's Schroedinger's cat. (Wikipedia that if you need to; it's interesting stuff.)

Yet we remain fixated on these single point forecasts. Why? I think, in part, it's a macho thing, and in part, it's a magical thinking thing. 

Michael Lewis' book Boomerang details how awfully off-base Iceland got in its financial sector, and he attributes a lot of the literally insane decisions that were made to the fact that the financial sector in Iceland was almost entirely dominated by men whose propensity to take ridiculous risks was far different from Icelandic women's approach to uncertainty. Studies of Wall Street prior to the financial meltdown have also shown that  macho risk-taking in a male-dominated business culture contributed a large, if ultimately unquantifiable, amount to the severity (and stupidity) of the 2008 disaster. Women-managed investment firms, according to Lewis' book, seem to have a more rational and safer record, investment-wise.

(One of my favorite movie quotes of all time was from The History Boys, where the female history teacher, Mrs. Lintott, tells her all-male charges, whom she is preparing for tests to get into Oxford and Cambridge, "History is a commentary on the various and continuing incapabilities of men. What is history? History is women following behind - with a bucket.")

So why do we continue to be drawn like moths to the flame of single point forecasts?

Folks, as a testosterone-poisoned male, I can tell you that there is NOTHING more satisfying than shooting at a target and hitting the bullseye. Talking about what I'm going to do if and when my arrow hits an outer ring, or misses the target completely, has little appeal. We males want to make a bold prediction and then hit the bullseye with it. In fact, the smaller the bullseye, the manlier we feel. 

Second, when we DO make a prediction and we DO hit the bullseye with it, we gain a certain prestige amongst our fellows from that, and again, the smaller the bullseye, the greater the godlike prestige. There seems to be something in human beings that responds viscerally to people who seem to possess some sort of mystical power to foretell the future.

Merlin was not a scenario planner. He was an arrow-shooter. 

Of course, unfortunately, he was also a myth.

But he's a powerful myth. We all want to believe in a free lunch, and in people with mystical powers to foretell. It lends a certain thrill to life to think that there are people out there who KNOW WHAT WILL BE.

And that, ironically, is what is selling Nate Silver's books - our fascination with the idea that someone out there must know the future.

The irony is that Nate himself is extremely realistic and modest about his ability to foretell. I don't think he would bet all his book royalties on his current 70.4% Obama forecast. Nor would he tell other humans to bet their homes on that outcome. I think he would recommend that we prepare for any outcome that has a non-trivial chance of occurring and a nontrivial impact on us.

And that means scenario planning.

But meanwhile, Nate's making a ton of money off our magical/macho fascination with his single-point forecasts.

I just hope that when he comes to figuring out what to do with that money, he's got a female investment advisor. 

Thoughts?