Thursday, May 25, 2006

Dumb Agent Theory

This is the theory that my thesis is based on. It is a theory that has been observed time and time again, although never proven per se. I am applying this theory to a future’s market, which I believe can be used as an indicator; in my case, for Foreign Direct Investment.
Right away I will say that this theory, being unproven, could be seen as rather tenuous to base my whole thesis on. It would be as if I had constructed an instrument capable of saving the Roman Empire based on the theory of Time travel. On the other hand, many occurrences, especially in physics, have been based on theory. The most prominent example is gravity, which comes about basically through process of elimination. Although this is all we have in theory, we have seen it in practice many times, and many other theories are based on it, such as observing the gravitational pull of stars in order to determine the position, size, mass and movements of planets that cannot be seen. That is why I treated this theory as such. Since it has been observed time and time again in many markets, and has never been shown to be incorrect, I assumed that it was correct.
Lately, however, I’ve been wondering if this is wise, so I decided to delve into the theory behind what happens to see if I can make sense of it.

The basis of any market, is the value of the underlying object or entity, or, more precisely, its utility. The difference between these is that an object could have an underlying value that is never utilized/observed/understood etc. I won’t get into whether something can be of value if it serves no purpose since that would bring me down a philosophical road I won’t know the first thing about. I will say that the value of any object or entity is its utility.

So, let’s take Object Alpha. Alpha has a utility, and therefore a value, of X. Can we guess what X is? If I were to guess I would probably say a number that is too high or too low. If any other person guesses, he or she will have the same results. How can we correct this? Well let’s take the other extreme: If everyone in the world guesses, values will be all across the board. But there will be a mid-point (or equilibrium point, or an amassed perceived value). This point must, by definition, equal Alpha’s value. If everyone estimates its value based on their perception of its utility, the amassed perceived utility must equal its real utility.

This follows along the lines that if I were the only person who could ever possibly achieved any utility (be it positive or negative) from Alpha, my perceived utility would indeed be its value. Therefore, everyone in the world together would be able to discover Alpha’s real value. So one person could achieve any value for Alpha, while everyone in the world would achieve its true value. In fact, everyone that gets any form of utility from Alpha would, collectively, be able to find its true value. Following this theory we can see why more liquidity means more preciseness in determining Alpha’s value, which basically proves the theory.

But what still bothers me is, if this theory can be shown to be true, why isn’t it used more often? I know companies have started using it, especially hi-tech companies, in order to determine what trends will last or pass away. How could one know ahead of time how big the I-pod would be, or how wireless keyboards would not pick up at all? It is also being used in order to predict outbreaks of influenza (I use this as a case study in my thesis) and has been shown to work well even with a very small number of traders. Maybe, like any idea, it just needs to have its time. When a market was proposed where people could trade on likelihood of a terrorist attack the Policy Analysis Market, the idea was immediately crushed in the house and the pentagon had to get rid of it. I still don’t understand why.

Anyway, if anyone has any thoughts, comments or constructive criticisms please let me know as I’d like to know this stuff backwards, forwards and sideways.


PhD Wannabe said...

I don't know if this helps, but skim through books by :

Bernard Salanie

and Jean-Jacques, Laffont

or stop by salanie office if you at columbia. think his office in sipa

danny said...

how is influenza related to FDI?

kim-ba said...

you've got good ole' fashioned logic here: (true A')'= true A. if you can't prove something false, it's true.

just a thought: you seem to minimize the importance of time and sequencing in the concept of 'real utility.' and what about volatility? or if the actual value of object x never reaches its theoretical 'real utility.' then what's the use of the dumb agent theory?

capitalizing off commodity x usually depends on predicting the changes in its value, right? same with political events, etc. ... so time and sequence are key. average utilities don't always calculate that. and if they sometimes do, how do you know when they do or don't?

Also, how do you know if your time frame is too small or too big to produce the real 'real' utility? Wouldn't you get different values if gathering the utility of roman empire from 500BC-1000AD versus the utility of roman empire from 10,000BC-2000AD timeframe?

Lugano said...

Danny: It isn't. At all. Except for the fact that it would be very useful if people were able to predict when and where it occurred and with what intensity. If these markets can be used to predict when and where Influenza will occur, and with what intensity, then there's no reason why we cannot do the same with levels of FDI into different countries. The fact that FDI is one of the best ways to measure the health of a country's economy is an entire separate point to itself, which I obviously also talk about in my thesis.

Lugano said...

Kim-ba: I think the whole discrepancy between utility and value could be a problem with this analysis. But, on the other hand, if I have a t-shirt, and let's assume no one else has any utility from it (no one sees it, misses it, took time to make it, loses anything or gains anything from it existing or not), and if I say it has a utility of $5, then its value is $5. I am the only possible market for it, so my perceived utility is its value. SO technically its value will, by definition, be its utility.

In terms of time, the Efficient Market Hypothesis (off of which the Dumb agent theory sprung about), says that any future value of anything is already incorporated into its value. This can change, obviously. But it will change as soon as new information comes about. For example the Roman empire might have had a perceived future value of a trillion in the year 200AD, but this value will have changed as the Northerners started invading, the Christians started gaining more power, and the Romans started partying too hard. Technically people who had grown up in the Roman empire for many generations might not think it would ever collapse, but if they had a futures market, its collapse would probably have been more and more foreseeable as these events occurred. So yeah, the Dumb agent theory cannot always predict the future, but it can show the real-time value of something before it is public knowledge using all the information at hand. And it will do this better than any individual person. There's an example of the Challenger disaster but that takes up space so I might write about it in another comment.

Lugano said...

Ok, what happened with the Challenger disaster was that, even though after it occurred NASA’s expert committee needed months in order to find the cause of this event, the markets gave their verdict immediately. Morton Thiokol, manufacturer of the faulty O-Rings, quickly received a 12% loss in share price, while the other shuttle contractors remained relatively free of market punishment.

That's it in a nutshell. You can find more info here:

PhD Wannabe said...

might be a useless comment for you might be done with your thesis by now..but anyway just want to say your examples are all random variables. Randomly occured with different levels of probability distribution. so maybe that's why any random kind of causes can predict any random effect. Example random seasonal rain could predict random stock prices given large enough samples.

PhD Wannabe said...

I assume your thesis study how people made decision (stock market, future market, fdi) under uncertainty. The rational way to predict your outcome is to use the expected utility and it is therefore based on the probability distribution on any outcome it could occur.

make any sense? or am i a fool again

BBS said...

let's say today you're going to sell something but realistically only about 10 to 20 people have the resources to buy all of what you're selling (they have to buy all). at the time of the initial purchase, they have to deposit a downpayment with you. if everything goes smoothly, the downpayment goes toward the purchase price.

after they purchase it (win by having the highest bid), they have about 2 weeks to resell 10% of what they bought to the general public (they can do whatever they want w/ the other 90%). all of the people bidding have about equal access to buyers when resold.

the catch is that they have to say today at what prices they will resell in 2 weeks time. if they are unable to keep the promise of reselling the 10% at the prices they indicate today, that constitutes default and they lose their downpayment which they paid for the privilege of bidding.

so, my question is, how do we determine the appropriate amount to charge for the downpayment?