firm_structure: Firm Restructuring at Regulatory Thresholds

This repository contains the replication package for the paper Firm Restructuring at Regulatory Thresholds. The study investigates whether threshold-based regulation leads to more fragmented firm structures.

---

Requirements

This code uses Orbis data (version: December 2024), accessible via the API provided by the TRR 266 Accounting for Transparency collaborative research center.

• If you are a TRR 266 member, please insert your API credentials by creating a firm_structure.env file. Use _firm_structure.env as a template.
• If you are not a member of TRR 266, it is advisable to access the required Orbis data via Moody’s SFTP delivery system.

A detailed list of required Orbis files can be found in:
code/01_pull_bvd_api_data.py

---

Setup

To run the code, you will need the following software installed:

• Python
• R

Required R Packages

The analysis relies on the following R packages:

• arrow
• data.table
• DBI
• dplyr
• duckdb
• extrafont
• fs
• future.apply
• ggplot2
• ggpubr
• glue
• modelsummary
• patchwork
• purrr
• scales
• tibble

---

Output and Workflow

To replicate the results, follow the scripts in the code/ folder in chronological order:

1. 01_pull_bvd_api_data.py
   Downloads the required Orbis data from the TRR 266 API.

2. 02a_prep_owner_data.R
   Constructs a panel dataset of owners from historic, time-static ownership files.

3. 02b_prep_financial_data.R
   Filters the financial data, following the approach of Beuselinck et al. (2023).

4. 03_sample_selection.R
   Selects the final sample, as described in the paper's sample selection section.

5. 04_aggregate_by_owner.R
   Aggregates firm-level statistics (e.g., total assets, employees) at the owner level.

6. 05a_descriptives.R
   Produces descriptive statistics and summary tables.

7. 05b_plot_regression_germany.R
   Generates the main regression results for the German setting.

8. 05c_disclosure_behavior.R
   Plots the share of firms disclosing income statements around the thresholds.

9. 05d_multiple_splits.R
   Identifies and presents case studies of owners who split firms multiple times to stay below thresholds.

10. 05e_cross_section_owner_named.R
    Tests whether owners of real estate firms are more likely to avoid thresholds.

11. 05f_cross_section_real_estate.R
    Tests whether owners of owner named firms are more likely to avoid thresholds.

12. 06_other_countries.R
    Investigates whether similar avoidance strategies are used in other countries.

---