Difference between revisions of "Information Asymmetry Measures"

From edegan.com
Jump to navigation Jump to search
imported>Ed
imported>Ed
Line 225: Line 225:
  
 
Match back to COMPUSTAT to get NAICS.
 
Match back to COMPUSTAT to get NAICS.
 +
 +
 +
===Accounting Variables===
 +
 +
Data source: COMPUSTAT - North America Fundamentals Annual
 +
 +
Ref vars:
 +
*Company Name
 +
*CUSIP
 +
*NAICS
 +
 +
Variables:
 +
*Market-to-Book-Assets (or Q): MKVALT (Sup: Market Value Total) / AT (Bal: Assets Total)
 +
*Market-to-Book-Equity: CEQ MKVALT (Sup: Market Value Total) / (Bal: Common/Ordinary Equity Total), TEQ  (Bal: Shareholder's Equity Total)
 +
*Earnings to Price Ratio: RE (Inc: Retained Earnings), EBIT (Inc: Earnings before Income Taxes), EPSPI (Inc: Earnings Per Share (Basic) Including Extraordinary Items), PRCC_F (Sup: Price Close - Annual - Fiscal)
 +
*Firm Size: AT, MKVALT (as above)
 +
*Development Stage (Sales<0.5b): SALE (Inc: Sales/Turnover (Net))
 +
*R&D Expenditure (XRD: Inc: Research and Development Expense), RDIP (Inc: In Process R&D Expense)
 +
*Ratio of R&D to Sales: XRD/SALE (as above)
 +
*Intangible Assets: INTAN (Intangible Assets - Total)
 +
*Ratio of Intangible Assets INTAN/AT
 +
*Sales Growth: <math>\frac{SALE_t - SALE_{t-1}}{SALE_{t-1}}</math>
 +
 +
 +
===Target Characteristics===
 +
 +
We already have:
 +
*Method of Payment
 +
*Diversification/Related (i.e., Horiz, vert, cong)
 +
*Acquirer Experience
 +
*Patent Counts
 +
 +
We need:
 +
*Distance btw Acquirer and Target
 +
*Citations Recd to patents
 +
 +
====Distance btw Acquirer and Target===
 +
 +
Addresses for both the Acquirer and the Target are available from SDC in the vast majority of cases.
 +
We will build a quick XML API to access: https://developers.google.com/maps/documentation/distancematrix/
 +
 +
We can pass 2,500 requests per IP address per day (24hrs) to this service, with a max of 100 elements per query and a max of 100 elements per 10 seconds. Note that URLs are restricted to 2048 characters, before URL encoding (particularly relevant if using multiple elements). Returned data includes:
 +
*Road Distances (in meters/ft in the value field, and miles/km in text)
 +
*Driving Times (in seconds in the value field, and hours in text)
 +
 +
Geocoding is internal. We can explicitly use Google's geocoding to get longitude/latitude if we want to calculate Great Circle Distances (etc): https://developers.google.com/maps/documentation/geocoding/#XML
 +
 +
Example query:
 +
http://maps.googleapis.com/maps/api/distancematrix/xml?origins=Bobcaygeon+ON|41.43206,-81.38992&destinations=Darling+Harbour+NSW+Australia&mode=driving&units=metric&sensor=false
 +
 +
The sensible thing is probably to query all targets (up to some limit, say 25, depending on the length of address strings) for a single acquirer in one-shot.

Revision as of 18:36, 26 July 2012

This page is referenced in:

  • VC Acquisitions Paper

This page provides a summary details from the Information Asymmetry in Acquisitions Lit Review and the begining of the build notes for these variables.

Table of Measure Usage

In a review of 28 papers that used one or more information asymmetry measures to explain stock price events (acquisitions, earning announcements, diversitures, etc.), the following measures were found:

D91 CS01 Mea07 KS99 FL04 O07 L92 T02 Aea02 AB94 LT07 T10 UC97 Oea09 M96 ES99 AL00 Y03 CS07 Cea04 Others* Count
Price/Volume Metrics y y y y y y y y 8
Idiosyncratic Volatility y y y y y 5
R-Squared from Earnings, Book Val. On Price y 1
Momentum y 1
Stock Illiquidity y 1
Pre-CAR y 1
Ratio of shares traded to outstanding y y 2
Abnormal/Unexpected Turnover y y 2
Analyst Forecasts y y y y y y y y y y 10
Forecast Error y y y y 4
Std. Dev. Of Forecasts y y y y y y y y 8
Normalized Forecast Error y y 2
Range of Forecasts y y 2
No. Estimates y 1
No. Analysts y y y 3
News y y y 3
News Announcements Y y y 3
Capital Structure y y y y y y 6
Breadth of Ownership/Block Holdings y y 2
Institutional Ownership/Holdings y y 2
Managerial Holdings y 1
No. Wholy Owned Subs. y 1
Accounting-based Measures y y y y y y y y y y 10
Market-to-Book-Assets (or Q) y y y y y y y 7
Market-to-Book-Equity y 1
Earnings to Price Ratio y 1
Firm Size y y 2
Development Stage (Sales<0.5b) y 1
R&D Expenditure y y 2
Ratio of R&D to Sales y 1
Intangible Assets y 1
Ratio of Intangible Assets y 1
Sales Growth y 1
External Responses y 1
Ratings Change y 1
Transaction Characteristics y y y y y y +8 14
Method of Payment y y y y y +7 12
Diversification/Related y y 2
Acquirer Experience y y 2
Distance between A&T +1 1
Target Characteristics y 1
Target Age y 1

Note that C98, Eea90, Aea90, BR91, Fea02, ET00, and CL87 (not listed above) all used payment method only, and BC11 used a distance measure.

Paper Codes

The paper codes are as follows (where 'ea' denotes et al.):

Code   Paper
AL00    Aboody and Lev 2000
Aea02   Affleck-Graves et al. 2002
Aea90   Amihud et al 1990
AB94    Atiase and Bamber 1994
BC11    Basu and Chevrier 2011
BR91    Brown and Ryngaert 1991
CL87    Calvet and Lefoll 1987
CS07    Capron and Shen 2007
Cea04   Carrow et al. 2004
C98     Chang 1998
CS01    Clarke and Shastri 2001
D91     Dierkens 1991
ET00    Eckbo and Thorburn 2000
Eea90   Eckbo et al 1990
ES99    Emery and Switzer 1999
FL04    Frankel and Li 2004
Fea02   Fuller et al. 2002
KS99    Krishnaswami and Subramaniam 1999
L92     Lee 1992
LT07    Lobo and Tung 1997
M96     Martin 1996
Mea07   Moeller et al. 2007
O07     Officer 2007
Oea09   Officer et al. 2009
T10     Tetlock 2010
T02     Thomas 2002
UC97    Utama and Cready 1997
Y03     Yook 2003

Summary Of Usage

There are 28 papers and 33 distinct variables broken into 8 distinct categories: Price/Vol, Analyst, News, CapX, Accounting, External, Transaction, and Target. Market Microstructure papers and measures are excluded from the summary and tabulation but included in Information Asymmetry in Acquisitions Lit Review with comments.

The average paper uses 2.8 variables from 1.7 categories. The most variables and categories covered by a single paper are 9 and 4 respectively, for Tetlock 2010.

Method of payment measures (cash vs. stock) are most popular, and present in half of all papers covered. Other transaction characteristics are rarely used. Acccounting-based measures, particularly Tobin's Q, and Analyst Forecast measures, particularly the Std. Dev. of Forecasts, are the next most popular and occur in approximately 1/3rd of all papers covered. Price/Volume measures, particularly the idiosyncratic volatility, are the next most popular, occuring in approximately 1/5th of all papers covered.

Rejecting Variables

The following variables are explicitly or implicitly related to the CAR in an acquisition and so unsuitable:

  • Pre-CAR
  • Abnormal/Unexpected Turnover
  • Momentum - It is unclear from a quick read of Tetlock (the only paper that uses this measure) whether momentum is an artifact of information asymmetry or a response to it's mitigation.
  • Stock Illiquidity - see Momentum above.

The following are simply too hard to get on a reasonable timescale:

  • News Announcements
  • Ratings Changes
  • Target Age (not in SDC, so we would have to get another source...)
  • R-Squared from Earnings, Book Value - this was used in a single paper and isn't worth it (it requires joining CRSP to COMPUSTAT before running the regressions)
  • All of the capital structure measures. Sources include 'Thompson 13' (Tetlock 2010), 'Value Line Investment Survey' (Emery and Switzer 1999), 'Compact Disclosure' data base of 13f filings (Utama and Cready 1997), 'CDA/Investnet' (Aboody and Lev 2000), etc. We don't have any of them...

Build Notes

Idiosyncratic Variability

We want [math]\sigma_{\epsilon}^2[/math] from [math]R_it = \alpha + \beta_i R_mt + \epsilon[/math] run annually for each publicly-traded firm in the NYSE/Nasdaq/Amex universe.

This is equivalent to the RMSE as:

[math]\mathbb{E}(\epsilon) = 0 \quad \mathbb{V}(\epsilon)= \mathbb{E} \left( (R-\hat{R}) - (\alpha-\hat{\alpha}) - (\beta - \hat{\beta})R_m \right)^2 = RMSE^2[/math]

Data:

  • Annual data from CRSP
  • Draw entire universe (>2Gb?)
  • Rely on date stamps
  • Use CRSP Permo (or Cusip?) - Don't need to draw NAICS if we are going to join back...
  • Holding Period Return
  • Value-Weighted Return inc. distributions

Run the regressions on raw data (i.e., don,t join to COMPUSTAT first).

Ratio of Shares Traded

Defined as: "the ratio of number of shares traded during the last year ending before the equity issue announcement, divided by the number of shares outstanding at the end of the fiscal year before the ... announcement."

We can compute it on an average over an annual basis using CRSP quarterly ("Shares Traded" isn't in COMPUSTAT), or using CRSP daily (same data as above), either way we want:

Data:

  • Share Volume (VOL)
  • No. of Shares Outstanding (SHROUT)
  • And to take an average over the year for each firm.

Analyst Forecasts

At least one paper reported problems with the data before 1991 (see the lit review).

  • Forecast Error (Analysts over and under react): [math]ForecastError=\frac{|ACT_t-EST_t|}{|Act_t|}[/math]
  • Std. Deviation of forecasts (Correlated with riskiness): [math] ForecastSD=\frac{SD_t}{|Act_t|}[/math]
  • Range of Forcasts
  • No. Estimates
  • No. Analysts
  • Normalized forecast error: [math]NormForecastError=\frac{ForecastError}{\sigma_{ACT_t - ACT_0}}[/math]

I.e., Over some time period calculate the detrended variation in Earnings. This probably isn't worth it. From KS99: "the normalized forecast error, which is defined as the ratio of the forecast error in earnings to the earnings volatility of the firm. Earnings volatility is the standard deviation of the firm's detrended quarterly earnings in the five-year period before the announcement of the spin-off."

Data:

  • From I/B/E/S Detail file pull:
  • CUSIP (8Dg)
  • EPS
  • Fiscal Yr1
  • Analyst Code
  • Estimate Value
  • Actual Value

Match back to COMPUSTAT to get NAICS.


Accounting Variables

Data source: COMPUSTAT - North America Fundamentals Annual

Ref vars:

  • Company Name
  • CUSIP
  • NAICS

Variables:

  • Market-to-Book-Assets (or Q): MKVALT (Sup: Market Value Total) / AT (Bal: Assets Total)
  • Market-to-Book-Equity: CEQ MKVALT (Sup: Market Value Total) / (Bal: Common/Ordinary Equity Total), TEQ (Bal: Shareholder's Equity Total)
  • Earnings to Price Ratio: RE (Inc: Retained Earnings), EBIT (Inc: Earnings before Income Taxes), EPSPI (Inc: Earnings Per Share (Basic) Including Extraordinary Items), PRCC_F (Sup: Price Close - Annual - Fiscal)
  • Firm Size: AT, MKVALT (as above)
  • Development Stage (Sales<0.5b): SALE (Inc: Sales/Turnover (Net))
  • R&D Expenditure (XRD: Inc: Research and Development Expense), RDIP (Inc: In Process R&D Expense)
  • Ratio of R&D to Sales: XRD/SALE (as above)
  • Intangible Assets: INTAN (Intangible Assets - Total)
  • Ratio of Intangible Assets INTAN/AT
  • Sales Growth: [math]\frac{SALE_t - SALE_{t-1}}{SALE_{t-1}}[/math]


Target Characteristics

We already have:

  • Method of Payment
  • Diversification/Related (i.e., Horiz, vert, cong)
  • Acquirer Experience
  • Patent Counts

We need:

  • Distance btw Acquirer and Target
  • Citations Recd to patents

=Distance btw Acquirer and Target

Addresses for both the Acquirer and the Target are available from SDC in the vast majority of cases. We will build a quick XML API to access: https://developers.google.com/maps/documentation/distancematrix/

We can pass 2,500 requests per IP address per day (24hrs) to this service, with a max of 100 elements per query and a max of 100 elements per 10 seconds. Note that URLs are restricted to 2048 characters, before URL encoding (particularly relevant if using multiple elements). Returned data includes:

  • Road Distances (in meters/ft in the value field, and miles/km in text)
  • Driving Times (in seconds in the value field, and hours in text)

Geocoding is internal. We can explicitly use Google's geocoding to get longitude/latitude if we want to calculate Great Circle Distances (etc): https://developers.google.com/maps/documentation/geocoding/#XML

Example query: http://maps.googleapis.com/maps/api/distancematrix/xml?origins=Bobcaygeon+ON%7C41.43206,-81.38992&destinations=Darling+Harbour+NSW+Australia&mode=driving&units=metric&sensor=false

The sensible thing is probably to query all targets (up to some limit, say 25, depending on the length of address strings) for a single acquirer in one-shot.