Analysis of Diff-in-Diff Data - Carrefour Rome

1 Introduction

In this hypothetical situation (the data is made up, i.e. does not exist), Carrefour Rome is evaluating the effectiveness of a coupon program designed to increase revenue by providing discount coupons for purchases over 40 euros. So they chose all the Carrefour Ipers and provided coupons from August until December.

2 Loading the Data

You can download the data from:

So, we want to calculate \(\beta_3\) from:

\[ Y_{it} = \beta_0 + \beta_1 \text{Group}_i + \beta_2 \text{Time}_t +\beta_3 (\text{Group}_i \times \text{Time}_t) + \epsilon_{it} \]

Question 1

Write the relevant equation for the problem and explain what the different terms represent.

Question 2

Create a DAG that maps out the variables that we care about.

Question 3

Create a static map for the area of interest and depict the different types of Carrerfour in different colors.

Question 4

Create a summary statistics table for the data and comment on the skewness or normality of the following variables: variety of products, store size, traffic, and revenue.

Question 5

What is the number of stores where the coupon program got implemented?

Question 6

What is the duration of the coupon program?

Question 7

Calculate \(\beta_3 (\text{Group}_i \times \text{Time}_t)\) from:

\[ Y_{it} = \beta_0 + \beta_1 \text{Group}_i + \beta_2 \text{Time}_t + \beta_3 (\text{Group}_i \times \text{Time}_t) + \epsilon_{it} \]

Question 8

Interpret the coefficients from the table.

Question 9

Create an event-study plot.

Question 10

Does the parallel trends assumption hold?

Question 11

What do you tell the Carrefour manager? Do the coupons work? Should Carrefour expand the coupon system?