- This repository contains codes for the following paper: Discrimination against foreigners in the U.S. patent system by Gaétan de Rassenfosse and Reza Hosseini, published in the Journal of International Business Policy.
In this paper, we looked at the potential discrimination by USPTO (United States Patent and Trademark Office) between foreign and domestic inventors. Inventions of foreign origin are about 10 percentage points less likely to be granted a U.S. patent than domestic inventions, controlling for invention quality. This finding adds to the body evidence that patent offices may be discriminating against foreigners, in apparent violation of international patent law. We show that this ‘bias’ against foreigners can be explained almost in full by differences in the quality of patent agents and in the financial resources of the applicants, as well as by the fact that domestic firms fight harder than foreign firms to get their patents granted.
Structure of This Repository
This repository has been organized into the following directories:
This folder contains the notebooks for generating the BigQuery tables from the CSV files for the PATSTAT tables.
This folder contains the notebooks for computing the custom family ID and creating the twin application's table.
This folder contains the notebooks and data files for combining different datasets and merging their information. Each notebook is used for creating a different set of features, which we then have used for creating the final dataset.
This folder contains the notebooks and data files for creating the final tables. The main notebook for preparing the tables is
Final_Dataset_Preparation. The steps for predicting names' ethnicity, country of origin, and gender can be also found in this folder.
Cloning the USPTO_2019 repository
cd destination/path git clone https://github.com/rezaho/uspto_2019.git
In case you need to run the BigQuery commands, you need to first set-up the Google Cloud console. Please follow the instruction in Google Big Query Quick Start.
We recommend you to create a new virtual environment using
Conda package manager.
The following code creates a new environment for Conda users using the
requirements.txt file provided in this repository:
# using Conda conda create --name <env_name> --file requirements.txt conda activate <env_name>
If you prefer installing the dependencies using
pip , you may use the following code:
# using pip pip install -r requirements.txt