Friday, May 01, 2009
On the virtues of economic modeling
by Larry Geller
Economic modeling is expository—it explains what is going on in the moment—and can have predictive value. You lay down a ruler or a curve, blanch, and kiss your kids’ college education fund goodbye.
I have a personal fondness for economic models of a different kind, the sort that attempt to build a machine in the computer that behaves like a real economy. You bang on it here, it goes in that direction. You pour money in here, and something happens that you can measure. You cut the budget here and see if the economy can stand the strain. The model is refined as it goes in an attempt to make it a better representation of reality.
My fondness comes from my youth as a college student and early computer programmer (gads, I sound so old…). ‘Way back in the late 1960’s (gads, I am so old…) I was a whiz at algorithms and programming GE’s timesharing computers. One gig I had was working Saturdays for a Merrill Lynch economist writing subroutines for his economic model. Weekend after weekend, subroutine after subroutine, I simply followed orders and coded them up. Then one day he put a few of them together… and a kind of crystal ball was born. By changing numbers in an input table, he got output. Then he tweaked and adjusted, and got better numbers.
I’m clueless to this day about what it was that I did exactly. I was carving pieces for a jigsaw puzzle that I never saw the whole of. All I know is that grinding out algorithms and subroutines paid my college expenses. For his company, the economist was able to demonstrate some things that were accurate for the present, and then he stepped back and let it run on into the future.
I am sure that others also ran his model. They didn’t have to know how it worked, just how to use it. That was what was cool about timesharing computers. You could sit down at a Model 33 TTY, call up a program, tell it a few things, and it would give you answers. It also played blackjack and had a cute Bullfight program where you selected cape moves and the program ran with your moves against the bull (intoning sometimes, “you are brave… stupid, but brave…”).
All three were models and were rather unique at the time. Remember, no one had personal computers really, except for the nuts with microprocessors, switches and no love life.
So the timesharing computer could model a game of blackjack, it had a crude model of a bullfight, and it had the economist’s sophisticated model of Wall Street. In explaining the economic model to people, I found I had to fall back on blackjack. They could grok that there were no real cards involved, it was all a simulation. For Merrill Lynch, there was no real money involved, it was also a simulation.
Hawaii could use an economic model. Then we could plug in an increase in the room tax and see how it affects room occupancy and the rest of the economy.
What will happen to unemployment if the room tax goes up five percent? If it goes down five percent?
The idea is to get away from blind accusations that any change is a good thing or a bad thing. Show me your evidence, or show me the model.
Same for restoring tax rates on upper income earners. Hawaii’s Republican Right abhors that, and they are entitled to abhor whatever they want to, but so what. Newspapers, most of them to the right of the general population, give them a sounding board. But it’s something of the bleat of the disappearing dinosaurs. Still, if there were a reliable model to plug into, we might have more confidence that government’s plans will bring about the desired improvements (for ordinary people, I mean). We could also model an economy transitioning from tourism to something else. That would be better than bleating from either side.
I wonder if there is such a thing going on somewhere in the state. If not, why not hire some college kid to help write one?