Authors:

  • Weimin Wang — Saint Louis University
  • Yong-Chul Shin — University of Massachusetts Boston
  • Bill B. Francis — Rensselaer Polytechnic Institute

Key findings

This paper studies whether trades by CFOs contain more information about future stock returns than trades by CEOs.

Main result:

CFO stock purchases are significantly more informative than CEO purchases.

Key numbers:

  • After open-market purchases:
    • CEO 12-month abnormal return: +2.41%
    • CFO 12-month abnormal return: +7.41%
    • Difference: +5.00%
  • Most excess return occurs:
    • Month 1–3: +2.58%
    • Month 4–6: +1.17%
    • Month 7–9: +1.02%
  • Even after public disclosure:
    • CFO still earns +6.19%
    • CEO only +1.57%
    • Spread remains +4.62%
  • CFO purchases predict stronger future earnings surprises:
    • CFO EA CAR: +0.430%
    • CEO EA CAR: +0.216%

Interpretation:

CFOs appear to possess and/or exploit more valuable information than CEOs when purchasing company stock.

Detailed notes

Research question

Traditional insider trading literature treats insiders as one homogeneous group.

This paper asks:

Do different executives possess different information advantages?

Focus specifically on:

  • CEO
  • CFO

Information quality is inferred through post-trade abnormal returns.

Economic intuition

Two mechanisms:

  1. Information advantage

CFO responsibilities:

  • financial reporting
  • budgeting
  • financing
  • capital allocation
  • cost control

Therefore CFOs may observe:

  • earnings quality
  • operating conditions
  • financial stress
  • accounting adjustments

earlier and more precisely.

  1. Incentive difference

Even if CEOs know the same information:

  • CEOs face more scrutiny
  • greater legal exposure
  • higher compensation outside trading
  • ownership guideline constraints

Therefore CEOs may trade less aggressively on private information.

Data

Sample:

  • Thomson Financial insider trading database
  • Jan 1992 – Jul 2002
  • Pre-SOX period

Final sample:

  • CEO purchases: 12,936
  • CFO purchases: 7,049
  • CEO sales: 24,527
  • CFO sales: 13,909

Filters:

  • stock price > $2
  • sufficient CRSP coverage
  • combine same insider same-day transactions

Methodology

Measure:

Post-trade cumulative abnormal return (CAR)

Benchmark:

  • Fama–French size × book-to-market portfolios (10×10)

Important technical contribution:

The paper avoids standard equal-weight bias.

Uses:

Return-weighted adjustment (Asparouhova et al., 2010)

Idea:

weight each return by:

\[1+r_{t-1}\]

to reduce microstructure bias.

Additional robustness:

Factor regressions:

  • CAPM
  • FF3
  • FF4 (momentum)

Core empirical results

Purchases dominate sales

Purchases:

  • strong alpha

Sales:

  • weak and mostly disappear after factor adjustment

Interpretation:

Selling is contaminated by:

  • liquidity
  • diversification
  • compensation liquidation

Buying is more information driven.

CFO advantage strongest in small firms

12M abnormal return:

Size Q1:

  • CFO − CEO = +5.83%

Size Q2:

  • +6.06%

Size Q3:

  • +5.48%

Large firms:

  • no meaningful advantage

Interpretation:

Information asymmetry is larger in smaller firms.

Market does not immediately absorb insider information

Even after SEC disclosure:

CFO purchases continue generating alpha.

Filing-day reaction:

  • CEO: +0.022%
  • CFO: +0.232%

Small relative to long-run returns.

Implication:

Market underreacts to insider signals.

Earnings explain only part of the effect

Earnings announcement CAR:

  • CFO purchase: +0.430%
  • benchmark: +0.150%

But:

earnings explain only a small fraction of total +7% return.

Meaning:

CFOs likely trade using broader information:

  • future cash flow
  • financing events
  • business conditions
  • strategic developments

Strengths

  • Clean executive-level decomposition of insider information
  • Strong economic magnitude
  • Multiple robustness checks
  • Links insider trading to future fundamentals

Weaknesses / limitations

  1. Pre-SOX sample only
    • disclosure rules changed materially after 2002
  2. Observational
    • cannot separate information access from incentive differences
  3. Long-horizon CAR methodology
    • sensitive to specification
  4. Alpha implementation ignores:
    • transaction costs
    • replication delay
    • modern disclosure latency

Suggestion on how to use the paper

Tag area: [alpha]-[insider-trading]

This paper is highly actionable for alpha research.

Ideas:

Insider hierarchy signal

Instead of:

\[\text{Net Insider Buy}\]

construct:

\[\text{Weighted Insider Buy} = w_{CEO} \cdot CEO + w_{CFO} \cdot CFO + w_{Other} \cdot Other\]

with:

\[w_{CFO}>w_{CEO}\]

CFO purchase intensity

Features:

  • purchase amount / market cap
  • recent purchase count
  • clustered purchases
  • CFO − CEO net activity

Combine with earnings revision

Since earnings only partially explains returns:

combine:

  • insider purchase
  • analyst revision
  • earnings momentum
  • accrual quality
  • quality factors

Event-driven implementation

Signal:

Recent CFO purchase

Holding:

3–9 months

Universe:

Small–mid cap

Controls:

  • size
  • momentum
  • liquidity
  • sector

Expected use:

medium-frequency equity alpha rather than HFT.

My takeaway

This paper is one of the cleaner examples showing:

not all insiders are equal.

For equity alpha research, executive identity itself can be treated as an information layer rather than using aggregate insider activity.