Description
This question will ask you to estimate the price elasticity of demand for fresh sardines across 56
ports located in 4 European countries with monthly data from 2013 to 2019. The data are
contained in the file EU_sardines.csv, which is available on Gauchospace.
Each row in the data file is a combination of port location (where the fish is landed and sold) in a
given year and month. You can ignore the fact that the sample is not balanced (the number of
monthly observations varies across ports).
For the assignment, you will need the following variables: year, month, country, port (port where
sardines are landed and sold), price_euro_kg (price per kg in €), and volume_sold_kg (quantity
of sardines sold in kg). In the questions below, I use log() to denote the natural logarithm.
(a) Estimate a bivariate regression of log(volume_sold_kg) on log(price euro_kg). What is the
price elasticity of demand for sardines? Test the null hypothesis that the price elasticity is equal
to -1.
(b) Like in Lecture 8 (see the IV.R script), we will use wind_m_s as an instrument for
log(price_euro_kg). To begin, estimate the first-stage regression relating log(price_euro_kg) to
wind_m_s. Interpret the estimated coefficient on wind speed. Does it have the expected sign?
Also test for the relevance of the instrument and whether it is a “weak” instrument by reporting
the proper F-statistic.
(c) Estimate the TSLS estimator of the price elasticity of demand for sardines using wind_m_s
as an instrument for log(price_euro_kg). What is the estimated price elasticity of demand for
sardines?
(d) Repeat the exercise in (c), but include fixed effects for each year, month, and country. [Hint:
you can use the command “as.factor(country) + as.factor(year) +as.factor(month)” to the ivreg
function in R]. Report the estimated price elasticity of demand and the F-statistic testing for
relevant and non-weak instruments.
