Replication Data for: Mapping Spatial Heterogeneity in Retail Advertising Effectiveness

We provide instructions, codes and datasets for replicating the paper by Luo and Ranjan (2025), "Mapping Spatial Heterogeneity in Retail Advertising Effectiveness".

The Code folder contains 4 R code files -

1. create_moments_for_synthetic_data.R : This file creates relevant moments for generating the synthetic data. This file requires access to the original datasets used in the paper (NDA-protected). The moments generated are saved in the “Data/Moments” folder. These are the moments required for subsequent Monte Carlo generation. This file is provided only to clarify the moment generation process, and cannot be sourced without access to the original data files.
2. create_synth_data.R : This file uses the moments generated from the above file to generate synthetic data. The data is first generated at the individual level and then aggregated to block-group level. Both datasets are saved in the “Data/Synthetic Data” folder.
3. summary_stats.R : This file loads the individual and aggregate level data and generates summary statistics tables and figures similar to Tables 1-3, Figures 2, 3 and 5 in the paper. Output is saved in the “Results” folder.
4. analysis.R: This file loads the aggregate level data and conducts empirical analyses to generate tables and figures similar to Tables 4-7, Figures 6-9 in the paper. Output is saved in the “Results” folder.

The Data folder contains the following files necessary for replication -
1. Moments/ : This folder contains the following moments generated from the original data that will be used for synthetic data generation
a. nind_quantiles.rdata : Empirical distribution of number of individuals sampled from each block group.
b. dist_quantiles.rdata : Empirical distribution of block groups’ distances-to-store for each retailer.
c. ad_stats.rdata : Models predicting price-promotional and non-promotional advertising levels
d. visits_model.rdata: Models predicting individuals’ visits to retailers as a function of absolute and relative proximity and fixed effects.

[A list of the versions of R, packages, and computer specification used in the paper]
R version: 4.4.0
Processor: AMD Ryzen 9 5900X 12-Core Processor
Installed RAM: 128 GB
To run the program, download all files (Code and Data) and save them under the same parent folder (and set the parent folder as the working directory in the code files). Then source the R files (create_synth_data.R, summary_stats.R, analysis.R) to run the programs.

[Technical Help or Problem Report]
We are expecting the R codes to be easy to use and to be working well. Please send any questions or report bugs to Bhoomija Ranjan (E-mail: [e-mail]).