Short Sellers and Financial Misconduct Jonathan Karpoff University - - PowerPoint PPT Presentation

Jonathan Karpoff University of Washington Xiaoxia Lou University of Delaware The Q Group April 5, 2011

Short Sellers and Financial Misconduct

Short sellers - April 5, 2011

Xerox’s cumulated market-adjusted returns January 1997 – December 2006

6/16/00: Xerox announces 2nd qtr 2000 earnings will not meet expectations 10/8/99: Xerox warns 3rd qtr 1999 earnings will be short of projections 7/3/00: SEC starts formal investigation 4/10-12/02: Wells Notice; SEC files civil complaint 3/26/07: SEC enforcement action concluded 1/1/97: Violation period begins

Short sellers - April 5, 2011

Public revelation

10/8/99: Xerox warns 3rd qtr 1999 earnings will be short of projections

Violation to revelation

Our questions

• 1. Do short sellers anticipate financial

misrepresentation?

• 2. How do they affect markets and social welfare?

Short sellers - April 5, 2011

Short selling is in the news…

A pernicious strategy of “short and distort … Market integrity is

threatened.”

Former SEC Chairman Christopher Cox

“Financial Terrorism”

CNBC’s Jim Cramer, on short selling

“Financial Jihad”

Infidel Blogger’s Alliance, on short selling

Short sellers - April 5, 2011

Defenders and detractors

Academics: Short selling facilitates information flow and the

price discovery process

Detractors: Short selling facilitates market manipulation

• Was Shamu actually ill?

Might short sellers be heroes?

• Jim Chanos helped expose Enron
• David Einhorn and Allied Capital

Short sellers - April 5, 2011

These views affect policy

• SEC actions during the crisis (September 2008):
• Banned naked short selling
• Temporarily banned all short selling in 799 “financial companies”
• New short selling reporting rules for institutional money managers
• Dodd-Frank bill (2010)
• Mandates new SEC study of short selling (to “protect” investor

confidence)

Short sellers - April 5, 2011

Our data

• All 632 SEC/DOJ enforcement actions initiated from 1988-2005 for

financial misrepresentation

13(b)(2)(A) - requires accurate books and records 13(b)(2)(B) - requires internal controls

• Monthly short interest data available for 474 firms
• CRSP data available for 454 firms

Short sellers - April 5, 2011

Timeline of a typical enforcement action

Violation Begins Violation Ends Violation Period Public revelation (“trigger event”) Inquiry Event Investigation Event Initial Regulatory Proceeding Concluding Regulatory Proceeding Regulatory Period Enforcement Events Proceedings Events

*

Violation to Revelation Wells Notice Enforcement Period

We focus on the time before public revelation

Short sellers - April 5, 2011

10 20 30 40 50 60 70 80 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

# of Cases Violation beginning to revelation (months)

Short sellers - April 5, 2011

Median time to revelation = 26 months

Num ber of enforcem ent actions and m onths to public revelation

Inquiry or investigation Regulatory proceedings

Down 18%

Violation Period

Down 3% Down 10%

Initial revelation

Initial Revelation

Short sellers - April 5, 2011

The Xerox example is typical…

Public revelation is bad news (Table II)

Question #1: Do short sellers anticipate financial misconduct?

Raw short interest = # shares short in month t

÷ # outstanding shares

Averages 1.65% over all firm-months

Short sellers - April 5, 2011

Abnormal short interest:

ABSI(j)it = raw SIit − expected SIit

Short sellers - April 5, 2011

Table III: Three measures of ABSI(j), j=1,2,3

Included in ABSI(2)

Short sellers - April 5, 2011

Table III: Three measures of ABSI(j), j=1,2,3

Included in ABSI(3)

Raw and abnormal short interest during months [-19, +20] (Table IV and Figure 2)

Short sellers - April 5, 2011

Raw short interest Abnormal short interest ABSI(1)

Is short selling related to the severity of the misconduct?

ABSI(1)i,-1 = 1.25 – 0.038 ARi + controls (p=0.03)

Abnormal short interest in month -1 Abnormal return upon revelation

But this could indicate merely that short sellers

anticipate a large price drop, not that they sell more when the misconduct is severe.

Short sellers - April 5, 2011

Three measures of misconduct severity

Fraud charges Insider trading charges Total accruals during misrepresentation period

Short sellers - April 5, 2011

From Table V

Inference: Abnormal short interest at month -1 is positively related to the severity of the misrepresentation – that is not yet public.

Short sellers - April 5, 2011

ABSI( j )

i ,−1 = λ 0 + λ 1Severity i ,−1 + λ 2Controls i ,−1 + ε i ,

j = 1,2,3

Panel A: ABSI(1) 1 2 3 4 Severity measures: Fraud charges 1.65 1.328 (0.03) (0.13) Insider trading charges 2.034 1.299 (0.01) (0.12) Total accruals 5.151 4.47 (0.00) (0.00)

The severity measures are economically meaningful

Change in severity measure Implied change in ABSI(1), in % Fraud 1.65 Insider trading 2.03 Total accruals: 10th to 90th percentile 3.02 Overall average level of short interest = 1.65% Average abnormal SI in month -1 = 1.9%

Short sellers - April 5, 2011

Fraud charges nearly double ABSI Insider trading charges more than double ABSI

Question #2: How do short sellers affect markets and social welfare?

Violation Begins Violation Ends Violation Period Public Revelation Violation to Revelation

Is this affected by short selling?

E.g., do they help uncover financial misconduct?

Short sellers - April 5, 2011

Table IX: Time-to-discovery logistic survival model

Models X(t) 1 2 3 4 5 Abnormal short interest

• 0.028
• 0.025
• 0.028
• 0.026
• 0.023

(<0.0001) (<0.0001) (<0.0001) (<0.0001) (<0.0001) Fraud

• 0.323
• 0.480

(<0.0001) (<0.0001) Insider Trading

• 0.008

0.122 (0.91) (0.13) Total accruals

• 0.228
• 0.197

(0.05) (0.07)

The amount of prior short selling is negatively related to the time until the misconduct is publicly revealed. Difference between 75th and 25th percentiles = 8 months

Short sellers - April 5, 2011

log(M i ) = β ' Xi + εi.

Do short sellers affect the price inflation during the violation period?

Violation to revelation

Would the price have been even higher without short selling? Price after public revelation

Short sellers - April 5, 2011

Actual price path

Short sellers’ external effects on share prices and quantity, for a given month t

How we estimate Phigh

retit = β0 + β1Sizei,t −1 + β2BTM i,t −1 + β3Momi,t −1 + Indik,t −1

k=1 K

∑

+ β4ABSI( j)i,t −1 + εi

retit

hyp = r it Ğö

β4 ABSI( j )

i,t−1

cumretit = retiτ

high τ =1 t

∑

Step 1: Cross-sectional model for each month t: Step 2: Hypothetical return in month t if abnormal short interest were zero: Step 3: Hypothetical cumulative return from the beginning of violation: Step 4: Phigh is the starting price * hypothetical cumulative return:

Short sellers - April 5, 2011

Data

External effects, mean estimates

Net benefit to uninformed investors = 1.09%

• f mkt cap.

Short sellers - April 5, 2011

Short sellers - April 5, 2011

External effects: Some details (Table X)

On average, short sellers convey substantial benefits to uninformed investors But for the median firm, these benefits are negligible (and the net benefit is slightly negative)

Summary:

The overall pattern of (abnormal) short selling

Early build-up Major build-up Wind-down positions Public revelation

The build-up is positively related to the severity of the misrepresentation

Short sellers - April 5, 2011

Early build-up Major build-up Wind-down positions Public revelation Short interest does not exacerbate the price decline when the misconduct is revealed

Short sellers - April 5, 2011

How do short sellers affect markets and social welfare?

Early build-up Major build-up Wind-down positions Public revelation Short interest decreases the time to public discovery of the misconduct

Short sellers - April 5, 2011

How do short sellers affect markets and social welfare?

Early build-up Major build-up Wind-down positions Public revelation

… And it dampens the price inflation during the violation period.

Short sellers - April 5, 2011

How do short sellers affect markets and social welfare?

Our questions answers

• 1. Do short sellers anticipate financial

misrepresentation? a) They anticipate the severity of the misconduct. b) They get it right, on average (low rate of false positives)

Short sellers - April 5, 2011

Yes.

Our questions answers

• 1. How do short sellers affect markets and social

welfare (at least with regard to this set of events)? a) They do not exacerbate price declines on bad news. b) They accelerate the discovery of the misrepresentation. c) They help keep prices in check while the books are in error.

Short sellers - April 5, 2011

Positively.

Short sellers - April 5, 2011

Appendix: Other issues examined

Do short sellers get it right in general? (Table VII) Does short selling exacerbate the price decline when

bad news hits the market (Table VIII)

Why does short selling tick up before the violation

period begins?

Other measures of severity (sizes of penalties

imposed)

Short sellers - April 5, 2011

The Xerox example is typical…

Public revelation is bad news (Table II)

Short sellers - April 5, 2011

Short sellers - April 5, 2011

Do short sellers focus on misrepresenting firms? (Table VII)

Only 1.78% of all firm-months are violation months But 4.18% of the “High short- interest” firm- months are in violation months

Public revelation

10/8/99: Xerox warns 3rd qtr 1999 earnings will be short of projections 4/10-12/02: Wells Notice; SEC files civil complaint 3/26/07: SEC enforcement action concluded 1/1/97: Violation period begins

Is this price drop larger when short interest is high?

Question #2(a): Do short sellers exacerbate price declines when bad news hits the market?

Or this drop?

Short sellers - April 5, 2011

Table VIII:

Models Variables 1 2 3 4 5 Abnormal short interest

• 0.314
• 0.231
• 0.185
• 0.191
• 0.052

(0.10) (0.23) (0.33) (0.34) (0.79) Fraud

• 9.610
• 8.528

(0.00) (0.00) Insider trading

• 11.72
• 10.52

(0.00) (0.00) Total accruals

• 5.963
• 2.810

(0.07) (0.37) N 339 339 339 273 273 Adj-R2 0.036 0.073 0.094 0.027 0.119 Controlling for the severity of the misconduct, short interest does not exacerbate the price drop.

Short sellers - April 5, 2011

ARi = a + f1 ABSI(1)i,-1 + f2 Severityi + f3 Controlsi + ei

Additional evidence:

When does the short selling begin?

Build-up before violation officially starts

• Does the violation start earlier than the SEC states?

Short sellers - April 5, 2011

Accruals and short interest at the violation start date

Low accruals at month before violation start date High accruals at month before violation start date …Implying that short sellers use accruals to help predict misrepresentation

Short sellers - April 5, 2011

Table IX, Panel B: Instrumental variable tests

Short sellers - April 5, 2011

ABSI( j)it = δ0 + δ1Optionsit + εi, j = 1,2,3.

Five ways to construct a sample of financial misconduct

• Our approach: All SEC/DOJ enforcement actions initiated from 1988-2004 for

financial misrepresentation

• 15 U.S.C. §§ 78m(b)(2)(A) - requires accurate books and records
• 15 U.S.C. §§ 78m(b)(2)(B) - requires internal controls
• Key word news (e.g., Lexis-Nexis) search
• Karpoff-Lott (JLE 1993); Murphy, Shrieves, Tibbs (JFQA 2009)
• Securities class action (10b-5) lawsuits (e.g., Gande-Lewis, JFQA 2009; Fich-

Shivdasani JFE 2007)

• 46% of our sample events have corresponding 10b-5 lawsuits
• Accounting and auditing enforcement release (AAER)
• AAER is a secondary designation assigned when the enforcement release names an

accountant or auditor

• Created in 1982
• May or may not have anything to do with financial misrepresentation
• GAO (2002, 2003) restatement database (e.g., Burns-Kedia JFE 2006)
• Currently popular

Short sellers - April 5, 2011

Universe of financial misrepresentations

Our sample

• Type I error (miss events in which

misrepresentation occurred) may be high

• Type II error (events include innocent firms or

individuals) is essentially zero

AAER samples

• AAERs miss 18% of enforcement actions, and

29% of all Administrative and Litigation Releases - so Type I error is higher

• AAERS include many instances in which there

is no financial misrepresentation (e.g., Boston Scientific – Securities Exchange Act Release 34-43183, also assigned AAER-1295) - so Type II error is non-trivial.

GAO restatement sample

• 1997–June 2002 (total = 919)
• Supposed to be cases of “fraud”, but SEC now says that

many are not misrepresentations or violations (implying that Type I and Type II error rates are very large)

Short sellers - April 5, 2011