EEP 142 Midterm Problems (Industrial Organization with Applications to Agriculture and Natural Resources)
1.(10 points, 2.5 each a) to d) )In a store, we displayed under the price tag whether the popcorn product was a low fat popcorn product. We were able to get data for before and after quantity sold for all popcorn products, in treatment and in two possible control stores. We also got data on the size of the popcorn display shelf for all three stores in our data set as shown below. (2.5 points each a) through d) below)
TABLE
Retail Low Fat Experiment: Summary Statistics
|
Treated Store |
Store C1 |
Store C2 |
||||||
|
Average Quantity before(*) |
110 |
100 |
230 |
|||||
|
Average Quantity After(*) |
130 |
160 |
300 |
|||||
|
Square footage of popcorn shelves |
150 |
149 |
260 |
The "Treatment Store" is the store where the intervention took place; the "Control Stores" are two nearby stores in the same retail chain, store C1 and store C2. Before period means weeks before experiment, after period means after experiment. (*) Average across weeks in Before and Average in Weeks in after period, respectively, average quantity sold is for all products sold e.g. week one 150 products are sold total, week 2 100, then average is 125 products).
a) Let C1 be the control. Give one reason as to why, given the table above, you choose that
one and not the other to measure the effect of the experiment.
b) Compute the effect of our nutritional labels. What do you conclude: do people appear to
like having a low fat label as measured by the responses in total popcorn sold in the
above table?
c) Suppose in Store C1 there was new rent-a-DVD display added during that same period
we ran the experiment. What would be the problem of using C1 as a control.
d) Given the problem in c) you have to use C2 as a control. How would you use the
information available to normalize the data and have somewhat comparable units?
Source | SS df Number of obs = 413590
-------------+------------------------------ F( 34,413555) =12909.53
Model | 4530767.84 34 Prob > F = 0.0000
Residual | 4268897.0 4413555 R-squared = 0.5149
-------------+------------------------------
Total | 8799664.89 413589
------------------------------------------------------------------------------
price | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
generic | -2.338719 .0114159 -204.86 0.000 -2.361094 -2.316344
class_id1 …
class_idN omitted
_cons | 7.312916 .0247798 295.12 0.000 7.264349 7.361484
a) What percent of the price variation in the data can be explained by generic characteristic and class type?
b) What is the average price in the dataset?
c) By how much are generics cheaper or more expensive on average?
d) Please test the null hypothesis that generic drugs are significantly less expensive.
e) If I explain price in terms of class, generic, and also in which stores it is sold at the R
squared of the regression changes to R-squared = 0.5193.
(0.5 points) Can you let me know what percent of the price variation can be explained by
location, i.e., explained by controlling for at which stores the OTC drugs are sold at (hint: relate
to the R squared in the above output)?
(0.5 points) Do you think we can conclude that there may not be large differences in prices
across stores?
-
Rating:
/5
Solution: EEP 142 Midterm Problems (Industrial Organization with Applications to Agriculture and Natural Resources)