Friday, September 23, 2011

Economists, Irrational Economies and Secret Assumptions

Here is a great video of Stephen Dubner ("Freakonomics") talking about the limits of social science experiments.  Specifically he showed that studies of charity change depending on how the question is framed.  That is, if you pay a college student $10 and then tell them they you have an extra $10 they can distribute among themselves and another anonymous person, they tend to give on average $3 to the other person and keep $7..  But if you tell them they can take give some of the money none, or even take  $1 from the base $10 from the other guy is supposed to get, then they tend to give $1.50. (Note, what happened was much more people gave nothing, it was not simply the people that gave $0 now taking $1.)

If you had not done the second two experiments, it looks like people are generous - they are giving $3 for no reason.   But if you allow the possibility of  'theft', the truth comes out.  Dubner stated that the charity was an an attempt to seem nice to the experimenter, and by 'not stealing' they look nice so they did not need to give.

My point here is that how you phrase the question dramatically affects how people react, and is part of the experiment.  It is not something you can 'write off' as irrational.

Back in 2010, Laurie Santos gave a talk about a similar issue  She compared monkey economies to humans (Video of a talk she gave) and declared the monkeys irrational for the choice they mad - which happened to be very similar to the choice people made in similar experiments.

In it, she failed to realize that the monkeys were smarter than the economists.   Specifically, she said that a brown capuchin monkey was irrational when it demonstrate loss aversion vs risk adverse, the same way that humans do.

The monkey version is comparing a choice of two scientists that displays 1 grape.  One guy half the time gave just the displayed grape and the other times gave 2 bonus grapes, for a total of 3.   The other guy always gave a single bonus grape (2 total).  In either case, the monkey averaged 2 grapes over the long term.   The monkeys usually took the safe bet of 2 grapes.  This is called risk adverse behavior.  The monkeys don't like to gamble.

But when you switch from the experimenters showing 1 grape (and giving extra) to instead showing 3 and sometimes taking them away, you get a different result.  In that case one guy always  took away 1 (leaving 2) while the other guy half the time took none (leaving 3), and the other half took 2 (leaving 1).  Here, the monkeys have radically different results.  Instead of being risk adverse, most monkeys are "loss adverse", preferring the risky potential of 3 grapes even though they might get stuck with just one.

Ms. Laurie Santos says this is irrational, as the monkeys treat loss different from gains, when in reality  the end result would be the same    In addition, she notes that humans act the same way the monkeys do  (If you switch out $1000 for the grapes, humans are more likely to gamble if they think they already own the money rather than if it is presented as a bonus).  She thinks this is an indication that humans have in built evolutionary flaws that screw with our ability to do economics.

But she is wrong.  The monkeys are smarter than she is.  

As Dubner said, how you present the choice affects the results.  It's not irrational, but affects us in logical, rational ways.   We are not trusting fools that believe what the experimenters say.   We live in the real world and we use clues from presentation to make our decisions.  These clues work.  I don't ride in cabs that reek of alcohol, even if the driver swears he is sober.   In social sciences, you are NOT measuring just what you want to measure, there are always a lot of extraneous variables.  In this case:
  1. The monkeys don't know the odds are the same.   They do NOT know that you have artificially set the chances for loss to be the same as the chances for gain.  Even assuming they can do the math and figure out that the odds are the same, it doesn't matter.  Why ?  See point 2 below.
  2. To quote every single mutual fund performance sheet: "Past performance is no guarantee
    of future results." 
    Things change.  People lie.   Ms. Santos, despite being well versed in monkey psychology does not realize this - at least she doesn't account for it.  The monkeys do not trust that the guys to keep acting the same way.
  3. The monkeys are not looking at the odds, they are looking at the MORAL CHARACTER.  They are not confronted with scientists offering risk/rewards scenarios.  They are confronted with humans - some of whom are nice and give 'tips', others of whom are nasty and steal.  
She did not set up an experiment where  the monkeys are calculating odds and rate of return.  Instead she set up an experiment where monkeys are deciding who to trust.
    Monkeys, like humans, treat tippers differently than we treat thieves.  When given a choice between someone that randomly tips well but sometimes tips nothing or a consistently average tipper, the monkeys (and I) would rather go with the guy that tips consistently average.  But that same rule does not apply to thieves.  Monkeys, like humans, would rather deal with someone that sometimes does not steal from them at all, even if when he does steal, he steals a lot, rather than the guy that always steals a little bit. 

    What does this have to do with politics?   It explains a lot of the hard to figure out stuff.   It is why the Republicans keep trying to say "Job Creators" rather than "Wealthy".   They tried to influence the voters.  It is why the Obama Health Care bill  described itself as "tax penalties for not having health insurance" rather than a tax hike that is exactly countered by a 'tax deduction for having health insurance'.  The Democrats did not want to admit they were doing a tax hike with an exactly equal tax deduction, because then the Republicans would have said TAX HIKE and left the rest out.

    But in a system where we make the rules, past performance may be no guarantee of future results, but it is a strong guide.  When it comes to taxes and economics, we need to look at what happened before.   As in Bush's tax cuts destroyed the surplus that Clinton created.   Look at the math  while spending has gone up, the truth is that the growth is relatively minor while the recession caused a huge drop in tax receipts.

    While character is important, it is not the only thing.  We need to look at the math as well as the intentions.

    No comments:

    Post a Comment