Friday, December 12, 2014

Updating the map to the territory

Well, I said I'd look into it.  But, actually, I don't think I have a reliable way to avoid adopting a vulgar pretense of principles.

Keynes famously said, "Practical men, who believe themselves to be quite exempt from any intellectual influence, are usually the slaves of some defunct economist."   This really doesn't only apply to economists.  I was reminded of this when I recently read this article about how John Muir's dismissive attitude toward Native Americans led to a lack of appreciation about how much some of the landscape he treasured had been managed by the indigenous people.  From this false premise, an entire paradigm toward ecology was built.  This article is what originally enticed me to write on this topic.  But, then, after finishing the first post, I have to admit that the analogy I sometimes use to compare economic policy to environmental policy utilizes this very premise.  Few people advocate meddling in the rainforest.  Yet, the forests we know have been deeply altered by human management for centuries.  So, I couldn't even complete the prequel to the post about intellectual honesty without Grubering it.*

What I was going to say, before I discredited myself, was that we all work out of some paradigm that consists of a set of interlocking presumptions.  Each of those presumptions, we hold with some confidence.  And, usually, if one presumption changes, it necessitates a change in the other presumptions.  The problem is that if you have a set of, say, 12 independent beliefs, and you have 80% confidence in each of them, you are 93% likely to be wrong about your broad worldview.  I am being wildly optimistic in saying that any of us holds less than 12 important beliefs that deserve an 80% confidence level.  If you think that you have a better set of priors than that, I can promise that, at least for a topic as complex as finance, we could easily find many smart people that would concede to you less than 50% on that many of your beliefs.

Most of the time, we operate in the least useful context for improving this situation.  We identify with groups of ideas or people that share a common set of beliefs.  When one belief is tested, which would create dissonance among other related beliefs, we pull back to more generalism.  We gather those other beliefs in defense of the belief that is being most directly tested.  One easy way to do this is through ad hominem.  We note that the source of the test has other beliefs that seem wrong, so that we can suspect that she has been misinformed in some way on the belief in question.  Or, with a little extra work, the same sort of discounting can be applied to the idea itself.  The fact that it contradicts several other beliefs creates doubt about the fact itself.

This is a very good reason to discount a new fact.  It is also a guaranteed method for avoiding any movement toward accurate beliefs.

Avoiding this is very difficult, because engaging in it is part of the essence of being social and being human.  I am frequently impressed by how Tyler Cowen can look at a source that might even treat him with contempt, and look for positive things to learn from it.  It's something I try to mimic, but ranting about others from within our own paradigms is so much more viscerally fun, our motivation is always biased toward self-satisfaction.

The first step to combating this is to insist that instead of pulling out to more generalized levels, we push to more detail.  We should look for ideas that test us and try to test the ideas on their own terms.  This is difficult to do in broad conversation with those who are vastly different than us, because most of our detailed experience will be informed by vastly different priors, so the details will generally seem preposterous to us.  For instance, in my previous post, Simon Wren-Lewis' statement that "the idea that private sector activity is always welfare enhancing and is best left alone was blown out of the water by the financial crisis" seemed like a ludicrous thing to say to me.  It would be like describing "Dirty Dancing" as a morality tale.  (Just goes to show you've got to clamp down on those teenage girls.)  But, in most audiences, I'm sure Wren-Lewis' statement is wholly uncontroversial.  Come to think of it, "Dirty Dancing" is a pretty effective morality tale.  There's some dark stuff going on down at the servant's quarters, and Baby is getting pulled in over her head.  But we all came in with the prior that you just don't put Baby in the corner, and priors rule our reaction to the movie.  We might hope that we wouldn't be a jerk-wad about it, but of course we would stop our daughter from sleeping with Patrick Swayze at summer camp if we could.  I mean, come on!

We take our reactions of outrage or incredulity as evidence of our own certitude, but frequently when we have extreme reactions, we are at our most unreliable.  Extreme reactions may even be biology's tool for giving us a way to fool ourselves.  My umbrage at Wren-Lewis' comment is a terrible reason to reject it, even if my umbrage is very convincing to me.

So, this should be our practice, considering ideas where there is some disconnect between our beliefs, or between theory and practice, or between our beliefs and other beliefs, and then dig down.  This is usually easier on the margins than with someone coming from an entirely different direction.  Usually, on my blog, I start with some sort of idea where a way of looking at something leads to a conclusion that is somehow counterintuitive to common treatment, and I'll follow it to where it goes.  I usually start writing before I have looked at the data, to help organize my thoughts.  Sometimes, like in this post on asset allocation, the data surprises me.  The way the post is written probably gives the impression that I knew what I was doing.  But, I had written the first half of the post, and was basically finished with a fairly standard description of the stock/bond relationship, when I thought, "Hmm, it wouldn't be that hard to check my data and see how close historical returns are to a normal distribution, just to double check this."  And, I saw that bonds have a terrible distribution of long term returns, so I wrote the second half of the post and went on to do a series where I decided that there was a relationship between stocks, real estate, debt, and bonds that I hadn't fully appreciated before.  I had noticed that the equity premium had not been historically consistent, but I hadn't considered the full ramifications.

But, surprisingly often, the data (or my interpretation of it) surprises me by agreeing with my hypothesis.  On the 11 part series of posts about leverage and risk trading, I had an idea about how leverage would effect profit margins in a counterintuitive way.  In thinking about it, it occurred to me that even though public discourse usually connects low interest rates to increased borrowing, the high margins I was trying to explain were coming from deleveraging in a low interest rate environment, and that even though this ran counter to common parlance, it was actually a corroboration of the Modigliani-Miller theory of taxed capital allocation.  Anyway, I wrote the first post before I looked at any data, because it was just a hypothetical. It took me a long time to pull all the models together after that first post, and I was shocked at how much the historical data seemed to confirm the theory, because it just doesn't seem like it should work that way.  (If true, it sure undermines any theory of the business cycle that assumes low bond rates are related to misallocated leveraged corporate investment.)

I also wasn't a housing bubble denier until I looked at the data.  I saw a provocative post suggesting that there wasn't an excess construction of homes in the 2000s, and as I thought about it, I started thinking of homes as securities.  (A friend tells me I turn everything into a bond.)  As I ran the numbers, and thought about oddities that don't fit into the common narrative of the housing market, I concluded that home prices, given the low real long term interest rates we have seen, were not significantly out of line.

But, these descriptions of my work are still just narratives I tell myself.  I've been rolling my eyes at Jonathon Gruber's transparent narratives, but in the end I don't know if his excuses are that different than my personal narratives.  When I look back at how striking the data was on corporate leverage, my memory is just as clouded and self-serving as Gruber's is when he tells Congress that he can't remember what he was thinking, but it must have been XYZ.  (XYZ happening to support a description of the law that would argue for Supreme Court support).  Maybe the raw data doesn't support Modigliani-Miller at all, but I'm as good at making narratives from the data as I am at making narratives about how objectively I interpreted it.

I would like to say that finance makes us much more honest than, say, politics, because our mistakes lead directly to personal costs.  But, a lot of financial analysis seems pretty flaky to me.  Of course there are still agency issues in many cases.  But, few things motivate us to backfill a narrative more than knowing we blew money on an investment or lost hard earned cash because of poor trading.  This is why I generally assume relative efficiency in arbitrageable markets, but in many cases I could (must) believe in Semi-Strong IMH (Inefficient Market Hypothesis).  That's where I trade.

Slate Star Codex had a great post on this problem, and how it complicates all contentious scientific endeavors.  In the end, it comes down to personal judgment.  Of course, as Feynman says,
"The first principle is that you must not fool yourself and you are the easiest person to fool."

The producers of "Dirty Dancing" could count on audiences universally leaving the theater saying, "Can you believe that mean dad tried to keep the street-wise itinerant dance instructor from sleeping with his naïve school-girl daughter?"  That's pretty strong evidence of semi-strong IMH (we are predictably, universally crazy in ways that we feel strongly about).  But, then, if my reaction to Wren-Lewis is "Can you believe that mean economist tried to keep Wall Street banks from betting your savings on risky securities?" the only reason I can claim that my reaction is reasonable is because I'm backing up to generalities and then populating the narrative with a bunch of my personal judgment calls that are a product of my paradigm itself.  Simply following different sets of seemingly reasonable priors very easily puts us in a position where even when we dig into the details and find something that clearly must be wrong, it will still be a product of the general.

But, it's the best we can do, and it has to be our goal.  Dig into those beliefs, and where we have 80% confidence, break down that confidence and test it, make that leg of our belief system fit a little awkwardly for a little bit.  And, when new information comes in regarding the other legs, we might see that a little change over there helps the new leg fit a little better.  Allowing a little dissonance to linger can lead to imperceptible paradigm shifts, until those shifts turn into a groundswell without us even noticing.  (And, before you know it, you're sitting before Congress wondering what you possibly could have meant when you said the things you said.  Which leads to another famous Keynes line: "If the facts change I change my mind, what do you do sir?" Even when looking back and seeing change in ourselves, can we tell the difference between growth and fecklessness, or between self-serving in-filling and brave readjustments?)

Where finance is helpful is that it relieves the pressure of trying to convince others or win arguments.  I really appreciate my readers who share their insights and reactions.  I started blogging partly to have those conversations.  But, if I'm working on a semi-strong IMH trade, I need the marginal investor to be on the other side. If semi-strong IMH is accurate, then the most persistently tradable incongruities arise from the truths that are right there to see, which we all disregard.  I can only really blog about my work as long as it mostly leaves other investors unmoved.  So, I'm trading notes with the world, but I don't necessarily want to convince you.  I want you to convince me that I'm wrong before I go putting money on the line.  Thinking that way really helps me to be more objective.  Not that I can say that I am objectively more objective than anyone else, just more objective than what I would have been if I was desperate to convince.

PS.  I apologize if these last two posts seem self-indulgent.  I promise posts will follow with charts and graphs.

*Gruber, v.: to promote a point of view by using the misconceptions of its detractors.

No comments:

Post a Comment