As I’ve been trying to come up with my own experiment for the final project, I found that is a remarkably difficult process to create a useful experiment that accurately portrays a scenario we would encounter in the real world. The problem, and this is a big problem, is that there is so much variance in the way people really go about trading. The last experiment we did provided some interesting results, but I don’t know how realistic of a scenario it was. Producers were limited to just two merchants—how often is this the case? Not very often, in fact it seems likely that there is an enormous pool of merchants who don’t produce anything themselves and who make their living solely by buying and selling goods for profit. That isn’t to say they aren’t useful, but only that they have much less bargaining power than they were endowed with in the experiment.
There is also a lot of trade that occurs only between producers. For instance, a company like YKK who manufactures zippers sells its wares only to other producers. There is no significant need for specialized merchants at all in this scenario, as the producers are acting as their own merchants. Although there was trade available between producers of the same village, it would be interesting to see what would happen if you opened it up to producers from other villages as well. I felt like this experiment created some artificial distinctions between producers and merchants—in reality there can be more of a gray area between the two. Especially with the advent of better transportation and communication the occupation of complete merchants is being demanded less and less. The position of merchant, largely, is being replaced by do-it-yourself producers.
There is also the problem that not everyone in a village will necessarily be a producer or a merchant. There will be plenty of people who are not in the labor force and who contribute very little to the productivity in the market, and there might be people who are producers but do not produce a good that is valuable anywhere but in their own village.
These are all nitpicky details, but they hold some of my frustration with trying to form an experiment of my own. There seems to be no fine line about where to say an experiment is “good enough.” Should experiments account for natural disasters? Should there randomly be a flood that destroys a producers output for every round? The level of detail you can place on these experiments is endless. Of course, the real question is would any of these changes have an effect on the outcome of the experiment. It’s hard to tell without actually testing all of the different possible scenarios. Some of the things discovered through the experiment, such as the casual nature of people in the same village versus the business attitude between the merchants, perhaps transcend any deviation from the experiment we participated in. After all, we tend to witness similar behavior in the real world—people are more cooperative with others they have a relationship with. Perhaps these are just laws that don’t depend on extraneous details.
But no experiment can hope to capture all the nuances of real life. By nature it has to be simplified in some form and condensed into a set of directions. Does this make experiments any less valuable? Not if the experiments can still create models that reflect the general principles we act on. We can assume that real life actions will not absolutely follow economic predictions with 100% accuracy, but we can extract motives and, hopefully, predict trends.
4 comments:
I think that the experiments we do still have validity even though they don’t necessarily take every aspect into perspective. The reason our environment was so limited in this experiment and in others is because we are testing one thing. We are testing how people respond to a certain stimulus and whether this response is predictable and statistically significant. You have to narrow down what you’re looking at before you can really see what you want to see.
Now I agree that this experiment was a very specific case, but this is how markets began. We didn’t have producers who only sell to producers and do it yourself producers to begin with. Instead, we originally had merchants and producers who specialized as a community. One community may create awesome bows and shields, another awesome shields and arrows and the last awesome bows and arrows. Now these communities need some way to trade in order to be efficient and that (on a very limited scale) is what we were testing through that experiment. The point of the experiment was to see how this type of market worked and if people would come to efficiency through cooperation and experimentation.
Every one of our experiments has been controlled in this manner. Even the double auction, which was almost true reality, was plagued by the promise that we would be able to sell our good again for exactly our value. This was a way to internalize an arbitrary feeling of value, but it still showed that supply and demand curves work in society.
Now regarding your experience with our final project, I had a very similar experience when trying to figure out my own experiment. I kept thinking there is no way to fit in every little part of human life into a computer experiment. I soon realized a few fallacies in my thinking however.
First off, the experiment doesn’t necessarily have to be on the computer. I originally only thought of the squares with little heads on top, walking around trying to complete some task I gave them. I soon realized that tests can be held anywhere and I decided to do mine with a rock concert and voting. You can do your experiment anywhere as long as you meet all the criteria and make sure you have a control.
I also kept thinking of great ideas but no questions to go with them. What we ended up doing was just looking through all of our questions we had turned in for the class and choosing the ones we thought would be interesting to look into and designing experiments for each. From there we chose the one we thought would be the best.
I hope I helped answer your questions regarding the limited view of experiments and our final presentation.
James, this is a good way to go about thinking of an experiment for your project—I like how you meticulously analyze the various aspects of our recent experiment that you found to be relevant or irrelevant in answering your own questions. After meeting with Prof. Wilson a few times to discuss my own project, one of the points of advice I came away was to focus more on a question that you are curious about rather than how to just simply create and experiment for the sake of it. That being said, while it is a good thing that you are asking questions about why certain aspects of the experiment may be more telling or realistic than others, it would be more useful for you to ask why a particular human behavior (such as defection) takes place—and then go about asking the minutely detailed questions in order to craft the parameters of your experiment to be 'good enough' to your own liking. In addition, I would like to address your belief that producers being forced to sell to only merchants is unrealistic and the comment from Mr. Pollard about its affect on the experiments validity and the origins of production. If your question has to do with the nature of decision-making processes' amongst a group(even a relatively small group of 15-20 people), and you believe that those processes' and consequently the experimental data can be drastically affected due to its rules being too rigid or unrealistic, then validity of the experiment is something that can in fact be tainted. The point is that even if you have a question and an experiment thought out, the results may be disingenuous or arbitrary if the parameters of your experiment are not in sync with the model you developed from your hypothesized question. For example, if your question has to do with the nature of trade amongst, say, producers and merchants vs. producers and producers, then it is important that you make the distinction that you did rather than rationalize the experiment's rules by identifying the similarities between the potential data-tainting controls and early mankind's predicament of narrow specialization as a result of relative isolation. Such a trivial similarity would matter little in validating the results if your question didn't pertain to such narrow parameters. In addition, just because markets didn't begin with producers selling to producers doesn't mean that there aren't identifiably unique behavioral traits amongst such producers—in fact such companies existed well before much of the 20th centuries established literature on behavioral economics had been written, such as Boeing and IBM. Given these facts James, it is important to 'nit-pick' as you have been doing since there is a fair chance there is a different set of behavioral traits that have evolved between producers for producers vs. producers for merchants.
In my own attempt to address your final comments on how you should tweak the rules of your experiment(once you have thought of your question of course), such as whether or not a flood should happen randomly, my guess is that you weren't struggling necessarily with how to implement your experiment physically, but how to find a system that can distinguish for you the difference between controls that will be either relevant or irrelevant in producing conclusive data. My suggestion is to remember that barometer you should be using to discover the validity of your experiment shouldn't be arbitrary, rather, it should be tailored to your ideas and your own vindicated empirical observations of market-place behavior. Once you feel that the rules you have set out for your experiment would produce the most realistic results by your standards relative to your question, then you should be in business. Good luck!
Post a Comment