Wednesday, September 15, 2010

Revealed Food Preference: A Tangled Web

Update: I originally posted this in March after participating in the experiment. I asked the authors for comments and they requested that I not publish until they were done running the experiment. Fair enough. I just attended the workshop on the basis of their results, so this can go back up.

Most of my concerns are not relevant for what they were trying to do but are still relevant for what we as economists claim to be able to do. (But what they *are* doing introduces a new set of issues I'll blog about after I read the paper and the methodology paper on which it's based.)
Time for me to demonstrate why they call it a degree in the PHilosophy of economics. In part, this is an esoteric problem that very few people should care about. In part, it's fundamental to what we as economists can say about the world based on the data we have available. Last week I learned that the relationship between what I buy at the store, what I would say on a survey are my food preferences, what I would tell a bunch of Econ101 students about my food demand, and what shows up in experimental data as my food preferences all give very different answers to my food preferences. Even if we are careful about definitions, it's a terrible mess.

For Euclidean geometry, three definitions are sacred and inviolable: point, line, and plane. Define those and a world of other proofs and theorems come from them. Change their definitions and you get a totally different geometry.

For economists, the fundamental element is preferences. A "rational" person can tell you between any two bundles which he or she prefers and the preferences over those bundles follow the property of transitivity. That is, if I prefer a pound of bananas to a pound of potatoes and I also prefer a pound of potatoes to a pound of tomatoes, then by golly, I really ought to prefer a pound of bananas to a pound of tomatoes. If not, I'm not rational.

It's very easy to be rational. Insane people can be rational. My professors have drilled into me that it takes a lot of hard work to not be rational. Appearances of irrationality can be deceiving. Even if you think you have a counter-example, we are very sophist-icated at demonstrating otherwise.

Last week I sat down in the experimental lab as a participant and by the end I realized the poor experimenters are going to be questioning my rationality.

They presented me and 21 other participants one pound of eight different fruits and vegetables in sequence. They did not show them all to us at once. We were to 'bid' for them by stating our "maximum willingness to pay" (WTP) for each food. Four of those 'auctions' would be binding in the sense that the person in the room with the highest WTP would actually buy that food. There is an important experimental economics question if that is a sufficiently binding rule to ensure truthful revelation - a question which we should seriously doubt depending on what data they care about from the 22 of us [they do]. But since I don't know which questions they were studying, I don't know. I only hope their results aren't sensitive to the folks in the lower range who know they have no chance of winning [they are; see future post].

Let's start at the beginning of my chain of reasoning as I go through this experiment.

1 - If they presented me with all 8 foods up front and asked me to rank order them according to how much I like eating each one, I could do that with no difficulty: Bananas > potatoes > grapes > carrots > red peppers > tomatoes > Empire apples > CA oranges. If they had presented me with Red Delicious apples instead of Empire apples, I would know how to rearrange the order, ranking the RD apples #3. My preferences are complete and transitive. I am rational. Oh good, I think, and breath a sigh of relief.

1.5 - They asked me this question, more or less, in the survey section. On a scale of 1-5 (hate to love), how much do you like each food? I can figure that one out too and translate my rank order into hate, dislike, indifferent, like, love. A slight permutation, but largely the same thing. It wouldn't have been more difficult to ask for a ranking of 8 than a ranking of 5, so why didn't they?  [Answer: They weren't thinking of it as a ranking, just a rating. It's a control variable that they aren't making much use of.]

2 - If they had asked me to envision my very own demand function for each of those foods, I could see in my mind's eye a possibly-kinked curve that intersects both the price and quantity axes for each food and is everywhere downward sloping. If they then asked me to follow those graphical representations of demand functions and find the maximum price for which I would be willing to purchase one pound of each food, I could do that too. I can readily imagine in such a world spending $6 (the maximum allowed bid) for a pound of bananas. They are second only to pizza in my food preferences, with chocolate and peanut butter a close 3rd. I might not do so very often, but I would buy that. My maximum WTP for a pound of bananas is at least $6.00

But they didn't ask me that. [They think they did.]

3 - You see, I and at least one other young man carried a reference or anchor price into the experiment with us. "I can go to Tops and pay $0.59 for a pound of bananas. Why on earth should I pay more than $0.60 for a pound of bananas?" He and I said that to each other after the experiment was over. My maximum WTP in that experiment was $0.60. The real world outside the experiment provides a ceiling: my maximum revealed WTP will be no higher than the store price, but could be lower if I don't buy any of that food.

Now, is my maximum WTP $6.00 or $0.60? Ouch, my brain hurts! I sat there philosophically debating this point with myself over and over again. If I told them $6.00, there was a 50/50 chance I would go home with a pound of bananas I would regret buying because I could have used the same money to get 10 pounds of bananas. I told them $0.60.

In fact, someone there actually paid $6.00 for a pound of grapes she could have bought at Tops for $2.00 - $2.50. Maybe she forgot her lunch and the convenience value of having a pound of grapes right then (the experiment was at lunch time) was at least $3.50. The guy I talked with could not believe her actions, which led to our conversation, but we did not expose her to our incredulous stares and whispers. I think it clear (but not necessarily certain) from her revealed preferences that she and we had a different understanding of the instructions and what a maximum WTP is. [Note: They claim no one paid more than $4 for anything. I could have overheard incorrectly.]

But it gets worse.
4 - Timing is everything. I'm on a diet. Everyone is perpetually on some diet or another. My [then] current lifestyle (South Beach Diet) does not allow me to eat white potatoes except on special occasions. I really like them (see rank ordering on stage 1). But every week for the past 9 months EXCEPT this one in preparation for my beloved St. Patrick's Day Dublin Coddle, I have walked out of the store without buying any potatoes. I had in my house right now all the potatoes I needed for the coddle.
Had they asked me 16 hours earlier, before I went shopping, what I was willing to pay for a pound of potatoes, I would have told them the store price. But because they asked me the day after, I had all the potatoes I wanted and I told them my maximum WTP was $0.00. [They reason that if people going to the store beforehand is uncorrelated with the advertising treatment, they're fine, and they're probably right.]

So on the survey I said that I prefer bananas to potatoes (love to like), and the price I'll pay for a food I love is more than what I'd pay for one I only like ($0.60 to $0.00). So far so good. But I'd better not be willing to pay a positive price for anything less than like, or my revealed preferences will start looking irrational.

5 - D'oh! They showed us advertisements for NY State apples interspersed with very old, poorly drawn, low-brow Simpsons cartoons. I don't normally go for Empire apples. They aren't Red Delicious, and neither is at their prime time of year. My apples NEED to be super crispy, and sweetness is definitely a plus. I would be willing to try their Empire apples to see how they compare to RD, but I would not be willing to pay store prices for them. Let's say $0.20 for my lowest ranked food in the hopes that it may be better than I expect. I could enjoy a crisp apple with lunch....


6 - It gets worse. I dislike tomatoes. I eat a little of them because it's good for me. But I am the primary shopper in our home and my wife loves tomatoes. She sometimes eats tomato sandwiches, or she did before South Beach took over. My utility is a function of her utility and our household decision making rules lead us to have 4-6 tomatoes on hand every week at $2.50 per pound.

In stage 2, I would have told them I'd pay at most $0.00 for a pound of tomatoes. I don't want them.

In stage 6, I told them I'd pay $2.50 for a pound that my wife would enjoy. I don't know what her stage 2 maximum WTP for tomatoes is. I do know that just yesterday she was praising the tomatoes I bought for $2.50/lb at the store.

So wait... My revealed maximum WTP is FOUR TIMES HIGHER for a food I don't like as for my second favorite food in the world?

My inner economist is weeping. How could we ever hope to disentangle household decision making rules, "household preferences" as the "aggregate" of individual preferences, individual preferences, demand functions, store purchase decisions, diets, quirks of timing, reference points, the effects of advertising, curiosity, and who knows how many other behavioral concerns? (e.g. The Lovely and Gracious reports that, had they showed her Simpsons cartoons, she would have been so off-put, she might not have been willing to buy anything from them at any price. "I don't know that $25 is worth watching a Simpsons cartoon" she says.) Yes, being more precise about definitions as I've tried to do here would help, but not much.

And this is microeconomics! Imagine the headaches we macro people ought to have.

Revealed value of my time for one hour spent questioning the very tenets of my professional faith: $25.Value of doing so while writing a book chapter on food consumption policies ... pric... can an economist believe that anything is really priceless? But that's a question for another post.

No comments:

Post a Comment