Post-Earnings Announcement Drift (PEAD)

An R pipeline to construct standardized earnings surprises (SUEs) for U.S. equity research via WRDS.

Overview

This repository constructs standardized earnings surprises (SUEs) for U.S. publicly traded firms, following the methodology of Livnat and Mendenhall (2006, JAR). The pipeline links IBES analyst forecasts with Compustat fundamentals and CRSP prices to produce three SUE measures per firm-quarter, which can be used to study the post-earnings announcement drift anomaly.

The code is an R port of the original Python notebook by Qingyi (Freda) Song Drechsler (WRDS, 2019/2021), updated to use the new CRSP v2 data tables.

Data Sources

All data are retrieved from WRDS (Wharton Research Data Services):

Source	Tables used
CRSP v2	crsp.dsf_v2, crsp.inddlyseriesdata, crsp.stocknames, crsp.msp500list
Compustat	comp.fundq, comp.security, crsp.ccmxpf_linktable
IBES	ibes.detu_epsus, ibes.actu_epsus, ibes.id

Sample period: January 2013 – December 2025.

Pipeline

CRSP–IBES link (get_iclink)
        ↓
CCM link table  →  GVKEY–Ticker–PERMNO universe
        ↓
IBES unadjusted estimates  →  latest estimate per analyst per firm-quarter
        ↓
IBES unadjusted actuals    →  merged on (ticker, fiscal period end)
        ↓
CRSP adjustment factors    →  adjust estimates to same per-share basis as actuals
        ↓
Compustat fundq            →  merge fundamentals and compute lagged EPS
        ↓
SUE calculations (sue1, sue2, sue3)
        ↓
data/esurprises.rds

SUE Measures

Variable	Description
sue1	(actual_EPS − expected_EPS) / (price / adj_factor) — no special items adjustment
sue2	Same as sue1 but with 65% of special items per share removed from both actual and expected EPS
sue3	(IBES actual − median analyst forecast) / price — analyst-expectation based

sue1 and sue2 are Compustat-based; sue3 is IBES-based.

Requirements

• R packages: tidyverse, lubridate, scales, RSQLite, dbplyr, RPostgres
• WRDS account with access to CRSP, Compustat, and IBES

Credentials

Set WRDS credentials as environment variables before running:

export WRDS_USER=your_username
export WRDS_PASSWORD=your_password

Or add them to your .Renviron file:

WRDS_USER=your_username
WRDS_PASSWORD=your_password

References

• Livnat, J., and Mendenhall, R. R. (2006). Comparing the post-earnings announcement drift for surprises calculated from analyst and time series forecasts. Journal of Accounting Research, 44(1), 177–205.