REQUIRED LINE OF BUSINESS REPORTING AND SHARE PRICE ACCURACY

1
REQUIRED LINE OF BUSINESS REPORTING AND SHARE
PRICE ACCURACY MERRITT B. FOX Joint Work With
ARTYOM DURNEV, RANDALL MORCK and BERNARD
YEUNG CELS Conference - NYU Law School November
10, 2007
2
LOB Regulations

• Required publicly traded issuers for the first
  time to disclose the sales and net income derived
  from each of the lines of business in which they
  were significantly involved
• Effective (for calendar fiscal year firms) with
  the 1970 10-k, filed in the first quarter of
  1971.
• Continues to be controversial

3
Question Studied

• What was the impact, if any, on share price
  accuracy and liquidity as a result of the SECs
  imposition of its line of business (LOB)
  regulations?

4
Approach

• We separately test the LOB Regulations impact
  by employing a number of measures used in
  empirical finance to study share price accuracy,
  information asymmetry and liquidity
• R², C², GAM, ILL, FERC, FINC
• Also used an Aggregate Index based on the first
  principal components of these other measures.

5
Data

• 400 NYSE and AMEX calendar fiscal year firms
  chosen randomly
• For around 363 (varying a bit depending on the
  measure), we were able to obtain on the CRSP
  tapes the needed price data for 1970 and 1971 as
  well as their 1970 Form 10-Ks
• Based on these 1970 10-Ks , we hand coded 209 as
  M firms (ones providing LOB disclosures) and 159
  as S firms (ones that did not)
• For each M and S firm, we calculated the value
  of each measure for the period before the
  imposition of the LOB Regulations and for the
  period after.

6
Difference in Differences Test Design for Each
Measure

• We compare the change for M firms in the
  measures value between the period before
  imposition of the regulations and the period
  after with the corresponding change in the
  measures value for the S firms.
• The regulations, by their terms, had no direct
  effect on the S firms.
• Thus comparison of the M and S firm changes in
  the value of the measure is between
• - a group of firms that were mandated to
  disclose more information than before, and
• - ones that were not mandated to disclose any
  more information than before.

7
R² Measure

• R² is based the extent to which share prices
  tend to move together in the same direction.
• A number of recent studies, both empirical and
  theoretical, associate reductions in R² with
  improvements in share price accuracy
• Thus R² has been used as a measure of share
  price accuracy

8
Findings R²

• R² decreases for M firms relative to S firms by
  a statistically significant amount

9
C² Measure

• C² based on intertemporal correlation of
  stock returns conditional on trading volume
• Theory
• with high trading volume dominated by traders
  motivated by private information, the periods
  returns tend to be followed the next period by
  returns moving further in same direction
• with high trading volume dominated by traders
  motivated by hedging, returns tend to reverse the
  next period
• C² is thus a measure of the amount of trading
  based on asymmetric information and,
  derivatively, an inverse measure of liquidity

10
C² Results

• C² decreases for M firms relative to S firms
  by a highly statistically significant amount

11
GAM (or Reversal) Measure

• GAM is based on the size of returns reversal in
  the period after there is large order flow
• Theory
• less liquid stocks would be more likely to
  over-react to order flow
• one would expect them to have larger returns
  reversal in the next period
• GAM is thus an inverse measure of liquidity

12
GAM Results

• No significant difference between the decline
  of M firm GAM and S firm GAM

13
ILL Measure

• ILL is a ratio that relates share price change
  to daily trading volume.
• lower ratio suggests share price is less
  sensitive to large buy and sell orders
• ILL is thus also a measure of liquidity

14
ILL Results

• ILL decreases for M firms relative to S firms
  by a statistically significant amount at the 10
  level

15
FERC and FINC Measures

• FERC and FINC for a firm with respect to a given
  time span is based on a regression of the firms
  stock return for each period within the time span
  on that periods change in earnings and on the
  change in earnings in one or more of the periods
  immediately following that period.
• FERC is based on the regressions estimated
  coefficients for the subsequent period earnings
• FINC is the difference in the explanatory power,
  as measure by r-squared, of the regression used
  to calculate FERC and a measure that only
  includes current earnings
• FERC and FINC are thus each measures of current
  share price changes to predict future earnings
  changes.
• Since future earnings are rough proxies for
  future cash flows, FERC and FINC are also
  measures of share price accuracy

16
FERC and FINC Results

• FERC increases for M firms relative to S firms
  by a statistically significant amount at the 10
  level
• No significant difference between the change in
  M firm FINC and S firm FINC

17
Aggregate Index

• The first principal components aggregate index
  is an orthogonal linear transformation of the
  data from the changes in each of the six
  preceding measures into a single dimension
• purpose is to extract common information from
  the six measures concerning the impact of the
  regulatory reform on share price accuracy and
  information asymmetry.
• the greatest variance by any projection of the
  data comes to lie on the first coordinate.
  (called the first principal component), the
  second greatest variance on the second
  coordinate, and so on
• an increase in share price accuracy and a
  decrease in information asymmetry would each tend
  to increase the aggregate index
• by construction, the average for the sample as a
  whole is 0.

18
Results of Aggregate Index

• The average for the M firms is negative and the
  average for the S firms is positive and the
  difference is statistically significant

19
Interpretation - LOB Regs Had Some Kind of Effect

• Taken together, these findings constitute very
  strong evidence that the LOB Regulations had
  effects on the pricing of the shares of the
  issuers to which they applied were priced
• The measures cut at the question in a variety of
  ways and in some cases have only a low
  correlation across the full population of firms
• 4 out of 6 of these measures showed
  statistically significant differences between the
  M firm change and the S firm change
• Aggregate Index also showed a significant
  difference
• Robustness checks on R² and C²
• Should settle debates about whether they
  prompted the disclosure of any meaningful new
  information

20
Interpretation - Liquidity and Information
Asymmetry

• Findings provide a substantial basis for
  believing that the effects of the LOB Regs
  included improved liquidity by reducing
  information asymmetry
• Difference in C² changes is highly statistically
  significant
• Difference in ILL changes is significant at the
  10 level

21
Interpretation - Share Price Accuracy

• Findings provide a substantial basis for also
  believing that these effects included increased
  share price accuracy
• Difference in R² changes is statistically
  significant
• Difference in FERC is significant at the 10
  level
• Further support comes from the evidence that LOB
  Regs reduced asymmetry
• could as a practical matter only happen by
  prompting the public disclosure of information on
  which insiders would otherwise trade
• information is reflected in price more quickly
  through public disclosure than insider trading

22
Implications for R² as a Measure of Share Price
Accuracy - R² Defined

• Run a regression on issuers return over time
• rj,t ß j,0 ß j,mrm,t ej,t,
• Decompose the total variance

23
Our R² Results are Driven by Lower SSM, Not
Higher SSE

• In most studies associating lower R² with
  increased share price accuracy, the SSM either
  contributed in a statistically significant way to
  the decline in R² or was not broken out
• Lower SSM driven decrease in R² is paired in our
  study with other measures indicating an increase
  in share price accuracy, which also suggests SSM
  driven decreases in R² can say something about
  share price accuracy
• Still most theorizing to date has focused on
  higher SSE driven decreases in R²

24
Noise Theory Story

• Noise traders, who trade on fads as fashions,
  must act in a herd fashion to have an effect, and
  it appears that they tend to do so across stocks,
  thus moving the whole market.
• Smart money speculators, who trade on the basis
  of fundamental information, engage in risk
  arbitrage that both combats the effect of noise
  traders and makes prices more accurate
• The extent of their effectiveness in doing so
  depends on the level of their activity
• Effective mandatory disclosure, by improving the
  improving the information environment and hence
  reducing their cost of information, increases the
  level of their activity

25
Noise Theory -contd

• Idea follows de Long et al (1990) relating to
  improvement in information environment for a
  country as a whole
• Improvement in information environment for M
  firms would represent an improvement for 60 of
  issuers

26
Noise Theory - contd

• The LOB Regs, if effective, narrowed the space
  for noise trading, but plausibly did so more for
  M firms than for S firms
• Market variance will decrease
• a systematic factor, the bias of noise, will
  play a less important role in setting the overall
  market return since, at a minimum, it plays a
  less important role in setting the prices of a
  significant proportion of all issuers.

27
Noise Theory - contd

• Narrowing the space will also reduce M firm
  betas relative to S firms
• each firms beta reflects its overall
  sensitivity to systematic factors that push all
  stocks in the same direction, including
• real economic factors, such as the interest rate
  or GDP growth
• noise trader biases also
• M firm share prices will become relatively less
  sensitive to noise trader biases and so their
  betas should drop relative to S firm betas
• SSM beta² x mkt variance
• Thus M firm SSMs and, derivatively R²s, will
  decline relative to those of S firms

28
Our Results Fit the Noise Theory Story Well

• Unusually large drop in overall market variance
  between 1970 and 1971
• 9th largest change in the 65 years between 1941
  and 2005
• Statistically significant drop, between 1970 and
  1091, in M firm beta relative to S firm beta
• 4th largest in 80 years

29
Implications for Role of Public versus Private
Information in Informing Share Prices

• Results run contrary to school of thought that
  share prices are primarily informed by private
  information
• The pairing of the reduction in information
  asymmetry and the increase in share price
  accuracy suggests that issuer public disclosures
  enhance share price accuracy
• more by directly altering the expectations of
  active traders
• than by promoting the generation of new private
  information
• Unless share price accuracy enhancements of
  public issuer disclosures is very small, which is
  unlikely because our tools for measuring them are
  not sensitive, this pairing suggests that the
  role
• of private information has been exaggerated