Experimental Evidence on Attitudes to Price Uncertainty

See attached .pdf. In short, we are playing experimental games with respondents in order to elicit their behavior in the face of (i) price risk, and (ii) price uncertainty.

2016-08-22

2016-09-30

The key outcomes of interest here are two. In the production game, the outcome of interest is the quantity produced by each subject in each round in response to market conditions (i.e., certain or uncertain price; if uncertain price, which of the four distributions the price will be drawn from). In the Holt-Laury lottery game, the outcome of interest is the subject’s switch point, i.e., the point at which they prefer lottery B to lottery A.

We aim for a sample size of n=96 subjects split into two equal groups of 48 subjects who will be playing price risk games and price ambiguity games. (See experimental protocols for a description of each type of game). The sample will be composed of farmers in rural areas around Lima, the capital. Those farmers have to be numerate and literate enough to play the experimental games. In order to make sure that they are, we present each would-be respondent with screening questions aiming at checking whether subjects have a basic knowledge of probability. If a would-be respondent fails that test, he or she is thanked for his or her time and not included in our sample; if the would-be respondent passes that test, he or she is then included in the sample.

In our experimental games, randomization will be done at the individual level using four-, six-, 10-, and 20-sided dice, so that there will be both within- and between-subject variation. In order to test that randomization was properly done, we will conduct tests using the ex-ante odds of being in each treatment for the entire sample. For example, in our first-stage randomization, the ex-ante odds of facing a certain price are 1/3, and the odds of facing an uncertain price are 2/3. When facing an uncertain price, the odds of being in each of our four price distribution are 1/4. So for the whole sample, we will test that the proportion of subject-rounds with a certain (uncertain) price are statistically equal to 1/3 (2/3). Then, within uncertain-price subject-rounds, we will test that the proportion of each scenario is statistically equal to 1/4.

Randomization done in office by a computer and the use of six-, 10-, and 20-sided dice during experimental games.

For the production game, randomization is at the individual-round level. For the lottery game, randomization is at the individual level.

No

0

96

n/a

n/a