Donald Trump the risk taker, confirmation bias edition

I am back after a holiday break — hopefully with fresh eyes through which to see US economic policy. Over the break, I got outside the beltway — to Miami, to London and to Cologne.  Germany was noteworthy because there was a lot of consternation over Americans’ electing Donald Trump to the presidency. At one German restaurant for example, when the server found out I was American, she lightheartedly — but seriously — took me to task for electing Trump.

There’s no comparison in German politics frankly, because no one who is not a career politician has a chance of winning in a parliamentary system of government. The closest thing to Trump in Europe is Berlusconi, who set up his own party in order to make it to Prime Minister in Italy. So what’s in store for us with Trump?

Look, here’s how I am going to go about this. I am going to put out a narrative on how I see Trump’s economic policy. My narrative is going to be that Trump is a risk taker. And the conclusion I have drawn is that this risk taking will lead to big surprises — some of them negative, but some positive. Afterwards, rather than give you a bunch of confirming information to back up my thesis, I’m going to tell you about my search for non-confirmatory data, because that’s how my process works. Only in the next post am I going to tell you why ‘Donald Trump the risk-taker’ matters and what it means for the economy.

See, a narrative is a story. And while stories are nice and they can be entertaining – especially when backed by data, human nature is such that people tend to create a narrative — a story —for others to hear, and then they search for the confirmation of that narrative, disregarding non-confirmatory evidence. This can lead to catastrophically wrong predictions.

What I try to do instead is create a narrative and specifically look for NON-confiming data – and tweak or adjust my narrative accordingly. Sometimes this means abandoning a narrative altogether. This process is ugly. Sometimes, it makes me look inconsistent. But, more often than not, I find it leads to better forecasts.

Let me give you a recent example of how this has worked in real life.

Remember when everyone told us Donald Trump was a joke and that Hillary Clinton was going to win the primaries and then the US presidential election? Nate Silver, for example, said Clinton had a 99% chance of winning the Michigan democratic primary. This was catastrophically wrong — because not only did she lose in Michigan to Bernie Sanders in the primary, she also lost to Donald Trump in the general election – ostensibly for similar reasons. And one reason she lost is clearer now than ever before. The narrative was wrong.

We were told to believe that a growing economy and low unemployment favored the standard-bearer candidate from the incumbent President’s party. And since Hillary Clinton was this person in this election, she was supposed to win the election. But she lost. And now that she has lost, people are looking for ways to reconcile the facts with their predictions – to figure out how they got it so wrong. People are looking to tweak the narrative after the fact. That’s a very different process than looking for non-confirmatory evidence in real time and tweaking the narrative along the way. In fact, this kind of ex-post narrative tweaking has led to the same kind of catastrophically bad confirmation bias that it did during the election cycle.

Did you know for instance that before George H.W. Bush won office for one term in 1988, no sitting Vice President had been elected President in over 150 years? That’s since Martin Van Buren in 1836. John C Breckinridge lost to Lincoln in 1960. Nixon lost to JFK in 1960. Hubert Humphrey lost to Nixon in 1968, and Al Gore lost to George W. Bush in 2000. And Nixon is the only former Vice President to win the presidency when he wasn’t the sitting VP. Henry Wallace and Walter Mondale both failed.

My point? Maybe, after a President serves multiple terms in office, the electorate wants change. Maybe Vice Presidents lose because they represent the status quo. That’s a narrative diametrically opposed to the prevailing one. And while I’m not saying I support that narrative, it is just as plausible as the one in which Hillary Clinton, as Barack Obama’s heir apparent, was going to win.

So here we are with Clinton losing to the now-incoming President Donald Trump, a man who has never served in public office, been a government official, in the military , or done anything as a government official. That’s definitely change.

But a lot of people are implicitly sticking with the old narrative though — the one in which Hillary Clinton was a slam dunk candidate. They think Clinton lost because of Russian “interference”. And there have been a lot of stories in the media about Russian interference as a result. Notice how the Russian interference angle implicitly confirms the old narrative though — i.e. if it weren’t for the Russians, the standard bearer of the President’s party should win during good economic times.

The problem, of course, is confirmation bias.

You’ve probably heard about that Vermont utility Russian hacking story. The original headline was “Russian hackers penetrated U.S. electricity grid through a utility in Vermont, U.S. officials say.” That is to say, the premise was that – in line with the narrative of Russian hacking costing the person who should have won the US election — the Russians are hacking US infrastructure too. This confirms the Russian threat and the Russian hacking narrative too.

The problem with this story? It never happened.

Instead, what we have since learned is that the malware-infected computer was not attached to the electric grid. In fact, the computer containing the alleged Russian malware also contained other malware too – something suggestive of a very different narrative i.e. that all of the malware on the computer was due to unsafe browsing or security habits of one of its users. Here’s how Forbes put it:

As many pointed out, the malware in question is actually available for purchase online, meaning anyone could have used it and its mere presence is not a guarantee of Russian government involvement. Moreover, a malware infection can come from many sources, including visiting malicious websites and thus the mere presence of malware on a laptop computer does not necessarily indicate that Russian government hackers launched a coordinated hacking campaign to penetrate that machine – the infection could have come from something as simple as an employee visiting an infected website on a work computer.

I am left thinking about this the way Bloomberg’s Leonid Bershidsky does:

It stands to reason that Russian intelligence was interested in the U.S. election campaign, and it’s a distinct possibility that it leaked what it found to the press via WikiLeaks, despite the latter’s denials. Russian President Vladimir Putin dislikes Hillary Clinton, and he probably would have been happy to hurt her chances of getting elected — thus, by default, helping Trump. It’s all quite logical, which is why a third of Americans believe Russia influenced the outcome of the election.

In the real world outside of soap operas and spy novels, however, any conclusions concerning the hackers’ identity, motives and goals need to be based on solid, demonstrable evidence. At this point, it’s inadequate. This is particularly unfortunate given that the DNC hacks were among the defining events of the raging propaganda wars of 2016.

And so, as I contemplate where Donald Trump is going to take his economic policy, the pitfalls of creating a narrative and seeking only confirmatory evidence is in the back of mind.

My narrative is that Donald Trump is a risk taker. And further, he is a risk taker who intends to take risks as President, on economic policy and on foreign policy. This creates a lot of uncertainty. And while there is upside from that uncertainty, there is also a downside as well. In the next post, I want to talk specifically about possible outcomes for trade policy.

Happy 2017.