WRDS

WRDS#

Wharton Research Data Services (WRDS) is a research platform that provides access to a broad range of financial, economic, and marketing datasets — including CRSP, Compustat, Execucomp, and many others. WRDS is operated by the Wharton School of the University of Pennsylvania; Kellogg maintains a site license that gives all faculty, PhD students, and approved researchers access.

WRDS is one of three platforms under Kellogg Data Hosting. Unlike Athena and Redivis, WRDS is operated by the Wharton School and accessed through their infrastructure directly.

Gaining Access#

  1. Go to wrds-www.wharton.upenn.edu and click Register.

  2. Register using your Northwestern email address (e.g., name@kellogg.northwestern.edu).

  3. Your account will be reviewed before access is granted — allow a few business days. If your work is time-sensitive, contact rs@kellogg.northwestern.edu .

Once approved, you can log in to the WRDS web interface to browse datasets and run small interactive queries.

Access Options#

Use the WRDS web interface to browse datasets and run small interactive queries. For programmatic analysis, connect from the Kellogg Linux Cluster (KLC):

Method

Best For

WRDS Web Interface

Browsing datasets, running ad-hoc queries, downloading small results

Python wrds package from KLC

Integrating WRDS data into Python workflows on KLC

R RPostgres package from KLC

Integrating WRDS data into R workflows on KLC

The rest of this page covers the programmatic Python and R options from KLC, which integrate directly into your research scripts.

Setting Up Credentials#

WRDS uses a PostgreSQL connection that requires a username and password. Rather than hardcoding credentials in scripts, save them to a .pgpass file in your home directory. Both the Python wrds package and R’s RPostgres read this file automatically.

On KLC, open a terminal and create the file with nano:

nano ~/.pgpass

Add this single line, replacing your-wrds-username and your-wrds-password with your actual credentials:

wrds-pgdata.wharton.upenn.edu:9737:wrds:your-wrds-username:your-wrds-password

Save and exit (Ctrl+O, then Ctrl+X), then restrict the file’s permissions:

chmod 600 ~/.pgpass

The chmod 600 step is required — PostgreSQL will ignore the file if it is readable by other users.

Setting Up Your Environment#

KRS maintains a shared conda environment on KLC with the wrds package pre-installed. Install the required R packages into your own R environment on KLC (only needed once).

module load mamba/24.3.0
source activate /kellogg/software/envs/wrds25_env

You can also install the package into your own environment:

mamba install -c conda-forge wrds
# or: pip install wrds

module load R/4.5.1
R -e 'install.packages(c("DBI", "RPostgres"))'

Connecting and Authenticating#

Open a connection to WRDS using your WRDS username. The .pgpass file supplies your password automatically.

import wrds

conn = wrds.Connection(wrds_username="your-wrds-username")

Test your connection:

print(len(conn.list_libraries()))

library(DBI)
library(RPostgres)

wrds <- dbConnect(
  Postgres(),
  host    = "wrds-pgdata.wharton.upenn.edu",
  port    = 9737,
  dbname  = "wrds",
  sslmode = "require",
  user    = "your-wrds-username"
)

Test your connection:

dbGetQuery(wrds, "SELECT 1")

Note

The first time you connect, WRDS will send a Duo two-factor authentication push to your registered device. Follow the prompts to approve it. Subsequent connections using the .pgpass file will not prompt for a password, but may still require Duo depending on your session. See WRDS two-factor authentication setup if you have not yet enrolled.

Exploring Libraries and Tables#

Before writing a query, you often need to know the exact library name (called a “schema” in PostgreSQL) and table name. WRDS organizes data into libraries by vendor; for example, CRSP data lives in the crsp library and Compustat data lives in comp.

You can also browse the WRDS Data Vendors page and WRDS Data Dictionaries in your browser.

List All Libraries#

A library refers to a database on WRDS (for example, crsp, comp for Compustat). Use conn.list_libraries() to retrieve a list of all databases your institution has access to. Sorting the results alphabetically makes them easier to browse.

libraries = conn.list_libraries()
print(len(libraries))
print(sorted(libraries)[:20])

libs <- dbGetQuery(
  wrds,
  "SELECT schema_name FROM information_schema.schemata ORDER BY schema_name"
)
print(libs)

List All Datasets in a Library#

Each library contains many individual tables or datasets (for example, company, funda, msf). Use conn.list_tables(library=...) to list all available tables within a given library. Specify the library name as a string (for example, 'comp' for Compustat).

print(conn.list_tables(library="comp"))

tables <- dbGetQuery(
  wrds,
  "SELECT table_name FROM information_schema.tables
   WHERE table_schema = 'comp'
   ORDER BY table_name"
)
print(tables)

Describe a Table#

Before querying, inspect the columns and data types of a specific table.

desc = conn.describe_table(library="crsp", table="dsf")
print(desc)

desc <- dbGetQuery(
  wrds,
  "SELECT column_name, data_type, is_nullable
   FROM information_schema.columns
   WHERE table_schema = 'crsp' AND table_name = 'dsf'
   ORDER BY ordinal_position"
)
print(desc)

For step-by-step workflows that pull and analyze data from specific datasets, see the worked examples below.

Worked Examples#

Note

The examples below are a selected subset of WRDS datasets with worked documentation on this site, not an exhaustive catalog. For an exhaustive list of datasets available to Kellogg researchers, see the Kellogg Research Support Datasets page .

Step-by-step examples for querying specific WRDS datasets from KLC. Each example covers a complete workflow in Python and R.

CRSP

Look up a stock by ticker, pull monthly returns from the CRSP monthly stock file, and plot cumulative return over time.

CRSP Example: Cumulative Return from Monthly Stock File
NYSE Trade and Quote (TAQ)

Fetch intraday trade data for a single stock on one trading day, compute five-minute average prices, and graph the result.

TAQ Example: Intraday Trades and Five-Minute Averages
Revelio Labs

Look up a firm by ticker, pull U.S. employment position spells, build a monthly weighted headcount and separation-rate panel, and plot the result.

Revelio Labs Example: Monthly Headcount and Separation Rate

A Note on Reproducibility#

WRDS datasets are updated and revised over time — data you pull today may differ from a pull made in six months. Save a local copy of any data you use in a project so your analysis is reproducible later:

df.to_csv("/kellogg/proj/your-netid/data/crsp_dsf_2023.csv", index=False)

write.csv(df, "/kellogg/proj/your-netid/data/crsp_dsf_2023.csv", row.names = FALSE)