Monday, May 14, 2012

Economists talk about utility maximization and profit maximization in the same way that these biologists talk about algae reducing drag coefficients...

Martone, Kost, and Boller (2011) write:

"Intertidal macroalgae must resist extreme hydrodynamic forces imposed by crashing waves. How does frond flexibility mitigate drag, and how does flexibility affect predictions of drag and dislodgement in the field?...  Bladed algae were generally “shape changers”, limiting drag by reducing drag coefficients, whereas the branched alga Calliarthron was an “area reducer”, limiting drag by reducing projected area in flow."

These guys don't actually think that bladed algae reduced drag coefficients. There are lots of equations in the article determining optimal drag coefficients, derived from the laws of fluid dynamics. Nobody is agruing that the algae get together, sketch out the optimal shapes they shift into at a given point on the blackboard given those equations, and then with perfect knowledge, foresight, and technique put it into practice.

But we do think there are good reasons to think the organisms optimize on critical margins. The systems in which they are optimizing may be complicated, but to the extent that we understand the systems they are in we can usually apply a little calculus and pretty quickly come up with this thing called "the optimal point" or "the optimal path" or "the optimal set". We always need to test those calculated optimals against what organisms actually do, of course, but since we've got a good reason (evolution by natural selection) to think organisms grope pretty consistently toward optimality, it's reasonable to calculate the optimality of the system and use it to talk about the behavior of an optimizing organism.

For some reason, people go nuts when mainstream economists do the exact same thing.

Of course human beings don't pull out pen and paper and figure out an optimal choice every time they make a decision. Nobody said they did. But we economists know a lot about the laws governing the social system that human beings operate in. We can bring a little calculus to those laws and quickly get a sense of the sort of behavior that a human being might engage in. We always take that to the data and test and change things where it needs changing. But generally speaking mainstream economics has held up pretty well (that's more or less how you survive as "the mainstream").

Figuring out how human beings approximate optimization - through heuristics, for example - is very important work, just like biologists' attempt to understand the approximation of optimization for animals generally is important work. We award Nobel Prizes for that stuff. We all support that stuff.

But despite the views of some, that is not a reason to dismiss the equally important work of understanding the often very complicated optimization issues involved in human social behavior, in order to develop testable predictions.

Think about the algae and their choice of drag coefficients next time you recycle some tired old criticism of "mainstream economics" and their boorish, naive mathematical models (and maybe try talking to a mainstream economist to see if they actually think the things you are accusing them of thinking, or if instead they're just trying to come up with a workable model of a very complicated system).


  1. I think part of the problem is that this may not be emphasized enough in economics college courses. My international relations courses made a big deal of the fact that our models were just tools to understand the world from the very first day of IA101. My economics courses felt like they were making a big deal that people are rational agents. Maybe it's just anecdotal, but then there is the fact that in many studies, econ students do behave much more rationally than their peers, happily violating social norms in the name of optimization. So I would venture to say that while those critics of mainstream economics may be wrong, mainstream economics has given them plenty of legitimate ammunition.

  2. "they're just trying to come up with a workable model of a very complicated system"

    policy setters attempting to "de-complicate" the system based on recommendations stemmed from mainstream economic models -- I imagine this is what pisses everyone off -- the reality of "uncertainty" and "entrepreneurship" by themselves should shine a light upon delusional recommendations stemmed from the notion of state enforced common good

    is it a tired old criticism to say you clearly don't have the tools in your tool bag to get the job done right (model/engineer society) -- no one does -- why should you be able to force people to care directly about your results -- telling individuals what they can and can't do verbally and or by decree is not how economics should be used -- do you think it is unfair that many view you guys as society's Pontius Pilate

    1. I'm not sure I understand what you're criticizing. Economists usually conclude with precisely this point - that you can't engineer society.

    2. Didn't you watch the Papola rap? You guys are top-down central planners.

  3. "But despite the views of some, that is not a reason to dismiss the equally important work of understanding the often very complicated optimization issues involved in human social behavior, in order to develop testable predictions."

    I agree right up to the part where you said "to develop testable predictions".

    Here is my test for how firmly one might buy into the "testable predictions" part:

    There is a perfect record of the historic price of gold. There is a perfect record of thousands of data points that varied in real time with the price of gold. With MATLAB, pick some dozens of variables and develop a polynomial that almost perfectly maps the historic price of gold with your chosen variables. Now, bet your your house that your polynomial will predict the price of gold tomorrow. Not a year from now, not a week from now. Tomorrow. You have ~80 years of history on prices and variables. And you are trying to predict the price 24 hours in advance. How can you miss?

    I am not saying that these models are not useful and insightful. I am saying that these models cannot be used to develop testable predictions any more than using fantasy football algorithms will tell us which teams will win next Sunday. When the predicted teams win, the researcher says "See?!?!", when the predicted team looses, the researcher says "I need to tweak my model."

    If you really thought math, when applied to economic variables, could predict the future, you would use math applied to economic variables to predict the future, with your own money. Some how, when it is one's own money, people become quite a bit less sure of the predictive ability of a given model.

    "Well, I mean if X, then generally speaking(variable of choice) will go up/down." Fine. How much money are you willing to bet on that specific variable? If the answer is 'none', why not? If you saw $1,000 in a bag the street, wouldn't you stop to pick it up? You KNOW what is going to happen. Or do you... (You not being you personally, but the generic you.)

  4. I think the difference is that there is only one variable that the algae adjust to here, whereas when humans make decisions there are a lot of different influences. You can say that *given* forces are changing, how do the algae respond? Economists say 'consumers generally maximise their preference rankings in the market' without really exploring the factors influencing this action.

    Also bear in mind that if this model produced a catastrophe comparable to the financial crisis it would be abandoned.

  5. First, evolution is a satisficer, not an optimizer. It's survival of the fit, not the fittest. Herbert Spencer did not get it right. In fact, some of the best evidence for evolution is the lack of optimality.

    Second, as Unlearningecon points out, the time scale matters. Generally in evolution the environment changes more slowly than the evolving organism, which gives the organism time to adapt. When the environment changes at the same rate as the organism, the result can be chaotic. Human systems in which agents are adapting to each other can be chaotic, as well.

    I have some other points I could bring up, but I am sympathetic. My main reason for posting is to recommend this link: :)

  6. So... you just happen to have an extracurricular interest in peer-reviewed botany journals, huh? "Hmm, what an interesting study I came across while browsing my morning RSS feeds. Wonder if I could segue this into a salient point about economics."

    I like the way you roll, Daniel.


All anonymous comments will be deleted. Consistent pseudonyms are fine.