My go to classroom experiment has been Veconlab's supply and opportunity cost experiment for at least a decade (here is a 2016 post). Here is the abstract from Holt et al. (2010):
This paper describes an individual choice experiment that can be used to teach students how to correctly account for opportunity costs in production decisions. Students play the role of producers that require a fuel input and an emissions permit for production. Given fixed market prices, they make production quantity decisions based on their costs. Permits have a constant price throughout the experiment. In one treatment, students have to purchase both a fuel input and an emissions permit for each production unit. In a second treatment, they receive permits for free and any unused permits are sold on their behalf at the permit price. If students correctly incorporate opportunity costs, they will have the same supply function in both treatments. This experiment motivates classroom discussion of opportunity costs and emission permit allocation under cap and trade schemes. The European Union Emissions Trading Scheme (EU ETS) provides a relevant example for classroom discussion, as industry earned significant “windfall profits” from free allocation of emissions permits in the early phases of the program.
I'm teaching an intro to environmental and resource economics class to mostly non-majors. For the third semester in a row I'm teaching online (n=45) and in-person (n=26). Both classes participated in the experiment on the second day of class. Here is the estimated linear probability model with Pr(sale=1) as the dependent variable, PRICE ranges from 1.5 to 8.5, COST is the marginal cost of production equal to 1, 3, 5 for units 1, 2, 3 in each of the 16 rounds, and GRAND is equal to 1 when permits are grandfathered and 0 when there is the permit price is $3. Through the magic of in-person teaching I stopped the experiment after four rounds of mostly ignoring opportunity cost and explained what was going on (DEBRIEF=1, 0 otherwise). The model is OLS with clustered standard errors (n = 20, t = 48 with the face-to-face data; n = 33, t = 48 with the online data):
Looking at the constants, the online class was more likely to sell the product with a 40% baseline probability (vs 26% in the F2F class). The F2F class put more weight on the price and cost. The price and cost coefficients should be equal since a $1 change in each effects profits the same way. But, in both classes more weight is placed on the cost variable. The F2F class ignored opportunity cost more than the online class. The debrief reduces much of the overproduction in the grandfathered rounds. But, overproduction is not much lower than for the online class after the debrief. The coefficient of determination (R2) tells me that the F2F class paid more attention to the variables than the online class.
Here is my attempt to turn this experiment into a journal article: https://www.env-econ.net/2023/09/new-working-paper-with-tanga-mohr.html.
