Friday, May 4, 2012

In defense of toy models

This is a quick note on the IS-LM model, and every other model out there that has been disparaged as being "too simple."  If it works, its complicated enough.  When you want to understand the world, your best bet is to use the simplest model that explains the wold you see, even if it's not strictly true.

I'm writing a paper on Argentina at the end of the 21st century, and while it is a stunningly complicated topic, the crisis can be cleanly explained with a straightforward classical IS-LM model (flexible prices and all), so long as you hold the nominal value of debt constant.  Were prices flexible in argentina?  Hell no!  It's a pseudo-socialist country with deeply broken markets and crony-capitalism type regulation.  But if you don't need to introduce sticky prices to explain the disaster, then why complicate things?

If this seems like a cop-out, lets talk science:

  • Gravity: 
    • in day to day life, we estimate it as "stuff falls down, heavier faster"
    • if you're calculating projectile motion, we calculate it as "stuff accelerates downward, 9.8 meters per second, per second"
    • if you're calculating planetary motion, you use the M*m*G/d^2
    • if you're anal about things, you use relativistic mass
    • if you really want to be right, you're SOL, since we still don't understand how this works
  • People:
    • walking in crowds, we use "people keep moving in a straight line"
    • dealing with strangers, we use "people follow cultural scripts"
    • dealing with business parters we use "people are trying to make a profit"
    • dealing with family we start working on understanding what people are thinking
    • if you really want to understand how people behave?  Yup, you're SOL.  Freud thought he knew, and that worked
I still have this paper to finish, so I won't go into more detail, but I'm sure you can.  We use the simplest model that works, and if you want to understand Argentina in 1999 (or spain today, for that matter) all you need is a simple IS-LM model where debts have a fixed nominal value.  So what if it's a "toy model?"  It works!

No comments:

Post a Comment