Do we know what we want Flashcards
In microeconomics, how do we find demand functions?
We assume people want to maxmise utility subject to a budget constraint which gives you a demand function, so assumes that consumers know what we want, but do they?
What is the paper related to this topic?
Coherent arbitrariness: ‘ Stable Demand curves without Stable Preferences’ - Ariely et al (2003)
What does the paper show?
COHERENT ARBITRARINESS - Experimenters show that consumers’ absolute valuation of experience goods is supringly arbitrary, even under full information conditions, but become imprinited ( coherent after initial decision is made.
What is the Becker-DeGroot-Marschak procedure?
The interviewer randomly assigns the respondent to one of two lotteries, in aim to elciit true willingness to pay.
4 steps
1) The interviewer randomly assigns the respondent to one of two lotteries.
2) The respondent is asked to imagine that they have won the lottey and are given a choice between two options:
1)Option A: Receiving a certain amount of money immediately
2) Option B: Receiving a smaller amount of money immediately and a larger amount of money at some later date.
3) The respondent is asked to choose between 2 options
4. The interviewer asks the respondent how much they would be willing to pay to receive the later payment in option B, so elicits true willingness to pay because it forces the respondent to make a real choice between two options and then quantify their preference
HOW MANY EXPERIMENTS ARE THERE
6
Experiment 1. There are 55 MBA students, who are shown 6 products e.g.chocolate, mouse book etc, the market price was not told.
Then subjects were asked to write down the last 2 digits of there social security number and then asked would you buy product x for $( the last 2 digits of your social security number, yes or no.
2) Then asked about WTP for the product in a incentive compatabile way ( if they are rational they state their exact willingness to pay) is done through the Becker-DeGroot-Marschak model? The table shows 5 groups with different bands of SS number e,g, quartile 1 would be those with SS number between 00-20, quartile 2 21-40 and so on.
With this in mind explain results?
Those with higher random social security number where willing to pay more for each of the products, this is because the question asked before anchors them to an acceptable amount to pay. This was sufficient to generate this pattern
The correlations row, shows the correlation of the random number and WTP for product and we see we have fairly large positive correlations that are significant.
What are the experiments 2-6 about?
We want to know people’s willingness to accept pain for exchange of payment.
The pain is an annoying noise
Then ‘ Would you be willing to hear it again in exchange for $X?
Then elicit WTA for different sound durations.
Why use sound?
Easy to sample.
No market so cannot base WTA on market price
Can repeat the experiment without altering experience.( sound does get better or worse )
Now lets look at experiment 2, in the first experiment the last 2 digits of the SS ( random variable) was an anchor for peoples WTP, we want to see if we can get similar anchoring effect here, also if we repeat the experiment does the anchoring effect persist. Finally are people consistent with choices e.g. if i listen to sound for longer, to they want to be compensated more?
So method
People get sample of the annoying noise, after that they are asked will they listen to it again, some are told for 10$ and others are told for 50$ ( hypothetical question, it has no bearing, so already an anchor)
After this subjects need to state WTA for listening to the sound again, there are 9 rounds of this. THE BDGM procedure used for people to elicit real WTA.
The graph below show the results, where there are 2 groups, 1st group get the sound in ascending order (10-30-60) and the other get it the other way around. So there are 9 rounds and they have to state how much they are WTA to listen to horrible sound.
Higher anchor = 50$
Low anchor = 10$
Remember there are 9 different sounds, hence WTA for the same time period are slightly different
What do results show?
Results show that WTA in the high anchor condition was significantly higher than average WTA in either the low-anchor condition or the no-anchor condition, WTA in low anchor not significantly different from no anchor condition.
Subjects are coherent, they want to be compensated more for listening to longer durations of the sound.
We also see that the lines don’t converge, so the anchoring effect persists over the 9 rounds.
How was the BDGM used for experiment 2?
TBA
Experiment 3, there might be a worry that people are using the monetary stated WTA first of all as an anchor, so the money holds signalling value, so to eliminate the possibility, we use the subjects SS number again, which is a random number, so it doesnt have any information.
They also use much higher stakes, so payoffs are much higher ( incentive to take experiment seriously )
They also have longer sound durations.
THEY ARE ALSO GOING TO CHECK WHETHER HIGH ANCHOR ( THEREFORE HIGH WTA) SUBJECTS SUFFER MORE FROM SOUNDS- shouldn’t be a difference. So determine whether high anchor group suffer more than low anchor group, they ask subjects to rate a number of unpleasant events including the sound and we compare this between the 2 groups and see no difference, thus those in the high anchor group who demanded more compensation didn’t do it because they suffered more.
So subjects now are asked to write the last 3 digits of SS number, then they are asked would they listen to the sound again for ( the SS number $)- so completely arbitrary.
So they split candidates into high and low anchor group depending on whether last 3 digits of SS number is below or above median.
What do these results suggest?
As we can see we get same results as experiment 2. So coherent arbitrariness persists even with randomly generated anchors and larger stakes
In experiment 4, slightly critiques the previous experiments, economists would say in real world, everyone knows the market price and be guided on it, however market prices are determined by peoples valuations.
Anyway in this experiment, they let people observe other peoples valuations, then do several rounds so people can learn what other people demand as compensation.
So your told this is the sound and you have to listen for this duration, you write down and submit your bid, the experimenter writes the bids on the board and the 3 people with the lowest bids will listen to sound and be paid the amount that is bid by the 4th lowest person, then get hypothetical question, would you be willing to listen again for X dollars. 9 trials.
Interpret the results.
findings were parallel with previous findings that lower anchor condition had lower average bids than the higher anchor condition. Even when people can see what the market price was, they persisted with high WTA with high anchor group ( EVEN IF THEY SAW BIDS WERE LOW) and LOW WTA FOR LOW ANCHOR GROUP
What was the BDGM procedure for experiment 5?
tba
Experiment 6, so previous experiments had demonstrated arbitrariness in money variations in money valuations. So does the anchoring effect only apply to money? so now they make the anchor Vinegar or gateraid. So they are made to either drink a certain amount of drink or listen to unpleasant sound What was the Modified BDGM method?
People make decision on whether to drink or listen to sound, once they switch this elicits there WTA.
So you make people choose and then tell them at the end that i will pick on the lines at random and you get what you chose at random, so its enough to elicit true WTA.
Interpret this KEY HERE IS THAT THE ANCHOR IS 1 OR 3 KEY HERE IS THAT WOULD YOU DRINK THE DIGUSTING DRINK OR LISTEN TO SOUND FOR SOUND FOR 1( LOW ANCHOR OR 3 MIN HIGH ANCHOR)
As you can see the WTA is significantly different for the 2 groups and is coherent, so it doesn’t only work when the anchor is money
So what is the final conclusion?
What is contingent valuation?
When you ask one person the question they will obvi be willing to pay more for london, but when you ask 2 different groups the same one of the questions the WTP are similar
