History
Before we get into anything else, some history. Because that stuff is important.
Daniel Kahneman and Amos Tversky - psychologists, not economists - are generally credited with the foundation of a myriad of related economics disciplines, including experimental economics. Their work and the subsequent development of Prospect Theory - a more nuanced version of the basic Expected Utility Theory of decision-making that was rooted in empirical results of certain experiments - is more usually linked with behavioral economics, but can also be taken as the beginning of experimental economics as a 'true' branch of economics.
Kahneman and Vernon L. Smith shared the 2002 Nobel Price in Economics, Kahneman for linking economics to psychology, and Smith specifically for the introduction of laboratory experiments as an approved method of creating and testing economic theories.
Generally speaking, a game is any controlled laboratory situation in which one or more participants are given a clear set of instructions and a specific set of potential actions that they can take in relation to those instructions. Participants 'play' with real money. No more context or explanation is given then is absolutely necessary for the rules of the game to be totally explained.
Seems pretty simple, right?
That's it.
So, what do you expect to happen?
If you immediately thought that the person with the money will keep everything, congratulations! You came up with what should happen, given the rules of classical economics and expected utility theory. Unfortunately, you also came to a conclusion that is exactly incorrect.
Specifically, according to this meta-analysis, in over 600 different studies of the game the average generosity of the dictator was about 30%. It was a left-skewed 30%, so dictators are more likely to give less than 30% than more, but still, this is pretty far away from people giving no money even a large portion of the time, right?
To be even more specific, only 6 of the 616 studies had an average 'give' of 0% and five of these occurred because the game was structured in a vastly different way than the 'normal' game (they had a 'take' option, rather than a 'give' option).
Daniel Kahneman and Amos Tversky - psychologists, not economists - are generally credited with the foundation of a myriad of related economics disciplines, including experimental economics. Their work and the subsequent development of Prospect Theory - a more nuanced version of the basic Expected Utility Theory of decision-making that was rooted in empirical results of certain experiments - is more usually linked with behavioral economics, but can also be taken as the beginning of experimental economics as a 'true' branch of economics.
Kahneman and Vernon L. Smith shared the 2002 Nobel Price in Economics, Kahneman for linking economics to psychology, and Smith specifically for the introduction of laboratory experiments as an approved method of creating and testing economic theories.
What's a Game?
Experimental economics uses laboratory experiments - hereafter referred to as games - as a basis for creating and testing economic hypotheses. So, what's a game?Generally speaking, a game is any controlled laboratory situation in which one or more participants are given a clear set of instructions and a specific set of potential actions that they can take in relation to those instructions. Participants 'play' with real money. No more context or explanation is given then is absolutely necessary for the rules of the game to be totally explained.
Seems pretty simple, right?
The Dictator Game
Here's an example of a game: take two people. Give one of them $20. Tell him he may give any amount of money to the other person that he desires.That's it.
So, what do you expect to happen?
If you immediately thought that the person with the money will keep everything, congratulations! You came up with what should happen, given the rules of classical economics and expected utility theory. Unfortunately, you also came to a conclusion that is exactly incorrect.
Specifically, according to this meta-analysis, in over 600 different studies of the game the average generosity of the dictator was about 30%. It was a left-skewed 30%, so dictators are more likely to give less than 30% than more, but still, this is pretty far away from people giving no money even a large portion of the time, right?
To be even more specific, only 6 of the 616 studies had an average 'give' of 0% and five of these occurred because the game was structured in a vastly different way than the 'normal' game (they had a 'take' option, rather than a 'give' option).
Variations
Perhaps more importantly, the meta-analysis we're looking at attempts to explain variations in giving between experiments by the differences inherit in those experiments.
Most important of these is demographics. Age, social status, and society all play a large role in the average give for an experiment.
For instance, average give tends to increase as age increases. Children and students are by far the most likely to give no money to the other person when they are the dictator, while elderly people are the least likely to give all the money to the other person.
People in developing and primitive societies are more likely to give more, as well. In fact, the average split when the game was played with participants from what could be described as a hunter-gatherer society was 50-50.
Other important factors include the size of the pot - more money means less giving, although the degree of the impact decreases dramatically as the amount of money increases - the framing of the situation - if the money has been 'earned' by the dictator they give less, while they give more if the non-dictator is somehow described or portrayed as 'deserving' - and various social factors - whether you know the person, whether or not giving is public or private, and the like.
Most important of these is demographics. Age, social status, and society all play a large role in the average give for an experiment.
For instance, average give tends to increase as age increases. Children and students are by far the most likely to give no money to the other person when they are the dictator, while elderly people are the least likely to give all the money to the other person.
People in developing and primitive societies are more likely to give more, as well. In fact, the average split when the game was played with participants from what could be described as a hunter-gatherer society was 50-50.
Other important factors include the size of the pot - more money means less giving, although the degree of the impact decreases dramatically as the amount of money increases - the framing of the situation - if the money has been 'earned' by the dictator they give less, while they give more if the non-dictator is somehow described or portrayed as 'deserving' - and various social factors - whether you know the person, whether or not giving is public or private, and the like.
Potential Explanations
There are a lot of potential explanations, many of which are not testable and others - such as the idea that people's ability to empathize with one another is the primary reason for giving - have been largely disproven.
For myself, I like to think that the reason money is given is because people tend to distribute the money in the way that most closely fits the way in which they believe it should be distributed. In this theory, you can treat the pot in the game as income that is given out perfectly randomly. People's responses, then, are as they would be if all income in real life were to be given out randomly, rather than through some combination of random factors and personal effort.
To test that, I'd like to see in the future a series of experiments that study the controlled effects of varying the randomness by which the pot is distributed, from 'pure' randomness (flipping a coin), to the current standard of unknown reasons for distribution, to varying degrees of the money being 'earned' (distribute by height, by weight, by how well you do on a quiz, etc.).
Could be interesting. I certainly think so.
So, next week and on Friday for the foreseeable future I'm probably going to be reviewing a different game that has been widely used in experimental economics. Next week will probably be the Ultimatum Game, which is a variant of the Dictator Game. It's interesting. Or, at least, I think it is.
For myself, I like to think that the reason money is given is because people tend to distribute the money in the way that most closely fits the way in which they believe it should be distributed. In this theory, you can treat the pot in the game as income that is given out perfectly randomly. People's responses, then, are as they would be if all income in real life were to be given out randomly, rather than through some combination of random factors and personal effort.
To test that, I'd like to see in the future a series of experiments that study the controlled effects of varying the randomness by which the pot is distributed, from 'pure' randomness (flipping a coin), to the current standard of unknown reasons for distribution, to varying degrees of the money being 'earned' (distribute by height, by weight, by how well you do on a quiz, etc.).
Could be interesting. I certainly think so.
So, next week and on Friday for the foreseeable future I'm probably going to be reviewing a different game that has been widely used in experimental economics. Next week will probably be the Ultimatum Game, which is a variant of the Dictator Game. It's interesting. Or, at least, I think it is.
No comments:
Post a Comment