Valuation of China

1
Valuation of Chinas Stock MarketMispricing of
Earnings Components

• In-Mu Haw Texas Christian University
• Shu-hsing Li National Taiwan University
• Donghui Wu Hong Kong Polytechnic University
• Woody Wu Chinese University of Hong Kong

2
Outline

• Introduction
• Literature Review
• Hypothesis
• Sample Data
• Empirical Tests
• Conclusion Implications

3
Introduction

• The development of Chinas stock market
• The stock market in China was established in
  early 1990s.
• Market capitalization ascended to the third
  place in Asia by April 2001
• Open to foreign money managers
• October 2002 first Sino-foreign fund management
  license.
• May 2003 QFII licenses are issued to Nomura
  Securities and UBS.

4
The number of stocks listed (1990 2007)
5
Total market value of A-shares(1990 2007)
6
The Shanghai A-Share Index(1990.12 2003.12)
7
Literature Review

• The rapid development of Chinas stock market has
  directed researchers attention to the role of
  accounting numbers in this market.
• Haw et al. (1999) Earnings are highly relevant
  to investors decisions.
• Chinese GAAP vs. IAS for the AB share firms
• Both Haw et al. (1998) and Abdel-khalik et al.
  (1999) IAS-based accounting numbers are not
  necessarily more useful to investors.
• Abdel-khalik et al.
• Can we make sense of the numbers?

8
Literature Review (Cont.)

• Underlying the above studies is the efficient
  market hypothesis (EMH).
• A convenient and parsimonious framework for
  understanding capital market.
• However, EMH may preclude us from discovering
  something that is not known by the market.
• There is mounting evidence suggesting that the
  market may not be as efficient as once believed.
• We put the descriptive validity of the market
  efficiency assumption in China to an empirical
  test.

9
Literature Review (Cont.)

• The mispricing literature
• The post-earnings-announcement drift
• Ball Brown, 1968.
• Bernard Thomas, 1989 1990 Ball Bartov,
  1996 Soffer Lys,1999.
• Pricing of earnings components
• Debt-equity swap gains Hand, 1990.
• Accruals vs. cash flows Sloan, 1996 Collins
  Hribar, 2000 Xie, 2001.
• Foreign earnings vs. domestic earnings Thomas,
  2000 2004.
• Special items Burgstahler et al., 2002.

10
Our Approach

• Decompose earnings into core earnings non-core
  earnings
• We find that
• Chinese investors do not differentiate core from
  non-core earnings.

11
Hypotheses

• Why decompose total earnings into core and
  non-core parts?
• According to the standard income statement
  prepared by Chinese firms (Figure 2)
• Core earnings (CE) operating net income
• Non-core earnings (NCE) all other I/S items
• Income from investments
• Government subsidy income
• Other items, e.g., gains or losses from disposal
  of fixed assets, assets revaluation, debt
  restructuring, etc.

12
Core earnings
13
Hypotheses (Cont.)

• In the valuation perspective
• ?CE are caused by changes in principle operations
• More likely to affect future operations
• Thus more persistent
• ?NCE are primarily caused by non-recurring
  transactions
• Less likely to persist into the future.

14
Hypotheses (Cont.)

• In the earnings management perspective
• To meet regulatory targets, Chinese firms often
  manage earnings by timing the transactions
  related to NCE (Chen and Yuan, 2004 Haw et al.,
  2005).
• Managed earnings are more likely to reverse in
  the next period.

15
Hypotheses (Cont.)

• Therefore, CE are expected to be more persistent
  than NCE.
• However, are Chinese investors aware of the
  difference between CE NCE and price them
  differently?

16
Hypotheses (Cont.)

• Why CE and NCE could be mispriced in China?
• The dominance of individual investors
• On 2002/10/31, at Shanghai Stock Exchange
• Individual investors Institutional investors
• 35,240,000 190,000
• 99.47 0.53
• High trading volume annual turnover rate gt 400.
• Is this justified, given Chinese listed firms
  the limited disclosures and low coverage by
  financial press?
• Is this driven by noisy traders?

17
Hypotheses (Cont.)

• If investors are unsophisticated and attach the
  same weight to CE and NCE, then
• CE are undervalued
• NCE are overvalued

18
Sample and Data

• Sample period 1995 2005.
• Sample firms
• All the firms listed in Shanghai and Shenzhen
  stock exchanges.
• 10,510 firm-year observations, representing 99.3
  of all non-financial observations during the
  period.
• Data source
• Financial statement CSMAR Genius
• Stock price and other data items CSMAR

19
Sample and Data (Cont.)

• Measurement of the earnings variables
• Core Earnings pre-tax earnings from principle
  operations.
• Non-Core Earnings all other income statement
  items.
• All the earnings variables are winsorized at the
  1st and 99th percentile.

20
Sample and Data (Cont.)

• Measurement of abnormal stock returns size- and
  BE/ME-adjusted returns.
• 55 benchmark portfolios are formed by sorting
  stocks into quintiles by their market value of
  equity and BE/ME at the beginning of each
  calendar month.
• Annual buy-and-hold abnormal returns ?(monthly
  raw returns mean returns of benchmark
  portfolios).
• This controls for the returns from
• rational pricing of the risk factors proxied by
  size and BE/ME, and/or
• mispricing associated with these two variables
  per se.

21
Monthly excess returns to the benchmark portfolios
22
The Actual and Implied Persistency of CE NCE

• The Mishkin (1983) framework for testing rational
  pricing
• Equation (1) estimates the actual persistency of
  earnings components.
• Equation (2) infers the persistency of earnings
  components implied by the market prices in year
  t1.

23
The Actual and Implied Persistency of CE NCE
(Cont.)

• Mishkin (1983) demonstrates that if the markets
  pricing of the value-relevant information is
  unbiased, then
• ai ai
• The consistency of regression coefficients
  between two equations can be tested by non-linear
  least square method.
• If ?2 (q) 2n Ln (SSRC/SSRU) is sufficiently
  large, then one can reject the rational pricing
  hypothesis.

24
The Actual and Implied Persistency of CE NCE
(Cont.)
25
The Actual and Implied Persistency of CE NCE
(Cont.)

• Therefore, CE is actually more persistent than
  NCE, which is consistent with
• NCEs transitory nature.
• NCE are more likely to result from earnings
  management.
• However, the weight assigned by investors to
• ?CE is significantly less than the actual
  persistence
• ?NCE is significantly greater than the actual
  persistence.

26
Predicting Future Stock Returns The Portfolio
Test

• If CE are undervalued relatively to its
  persistency, then the market would be surprised
  by the higher earnings realization in the
  subsequent period.
• ?CE should be positively related to next periods
  returns.
• Similarly,
• ?NCE should be negatively related to next
  periods returns.
• Therefore, profitable portfolios can be formed by
  the earnings composition.

27
Predicting Future Stock Returns The Portfolio
Test (Cont.)

• ?CE Portfolios
• Long (short) in the stocks with largest
  (smallest) ?CEt within each ?Et decile.
• ?NCE Portfolios
• Long (short) in the stocks with smallest (largest
  ) ?NCEt within each ?Et decile.
• Changes in total earnings (?E) is controlled for
  in the above strategies.
• ?CE?NCE Portfolios
• Stocks are first grouped into quintiles by ?CEt
  and ?NCEt independently.
• Long (short) in the stocks that are in both the
  top (bottom) ?CEt quintile and bottom (top) ?NCEt
  quintile.
• Information on both the ?CE and ?NCE is utilized
  simultaneously.

28
Predicting Future Stock Returns The Portfolio
Test (Cont.)

• The portfolios are formed at the beginning of May
  after year t and held for a year.
• Implementable trading rule
• Predictive test.
• While firm size and BE/ME effects are controlled
  for in measuring abnormal returns, some unknown
  risk factors may still lead to positive hedge
  portfolio returns.
• Therefore, we report yearly abnormal returns on
  the portfolios and infer statistical significance
  by
• T-test based on time-series variations of
  returns.
• Binomial test based on the signs of the yearly
  returns.

29
Predicting Future Stock Returns The Portfolio
Test (Cont.)
30
Predicting Future Stock Returns The Portfolio
Test (Cont.)

• The positive returns to the hedge portfolios
  suggest that information on current earnings
  composition predicts future stock returns.
• Thus, current information is not fully reflected
  into stock prices when it is available.
• The returns are positive for most of the sample
  years.
• It would be difficult to attribute the returns to
  some unidentified risk factors.

31
Predicting Future Stock Returns The Portfolio
Test (Cont.)

• The concentration of abnormal returns during
  earnings announcement periods.
• Concentration would occur if a large amount of
  unexpected earnings information becomes available
  to market participants on the earnings
  announcement dates.
• If abnormal returns are simply risk premium, then
  the higher or lower returns should be evenly
  distributed in year t1.
• Earnings announcement periods include the 3-day
  windows centering on the interim and annual
  earnings announcement dates.

32
Predicting Future Stock Returns The Portfolio
Test (Cont.)
33
Predicting Future Stock Returns The Portfolio
Test (Cont.)

• Why there is presence of clustering of abnormal
  returns in the short positions but absence of
  clustering in long positions?
• Low litigation risks against the Chinese managers
  firms are likely to encourage early disclosure
  of good news.
• Information contained in good news earnings
  announcements is more likely to be preempted by
  other sources than that in bad news
  announcements. That is, bad news travels slowly.

34
Predicting Future Stock Returns The Regression
Analysis

• The regression approach
• More convenient to control for other factors
  affecting both the stock returns and our
  experimental variables.
• Test whether the mispricing of CE is incremental
  to NCE.
• The regression model

35
Predicting Future Stock Returns The Regression
Analysis (Cont.)

• Cumulative raw returns are dep. var., and two
  market return indexes are indep. var.
• Firm size and BE/ME are used as indep. var. to
  control for normal returns.
• Seasoned equity offerings (ROt and ROt-1),
  closeness to delisting (Delistt), and modified
  audit opinions (MAOt).

36
Predicting Future Stock Returns The Regression
Analysis (Cont.)
37
Predicting Future Stock Returns The Regression
Analysis (Cont.)

• The regression results are consistent with those
  from portfolio tests.
• Furthermore, the regression analysis suggests
  that mispricing of CE is incremental to that of
  NCE.

38
Effect of Delayed Responses on Value Relevance of
CE NCE

• The value relevance of earnings how relevant
  are earnings to users pricing decisions.
• Earnings response coefficients (ERCs)
• RETt a ßEARNt e
• ERCs ß, the rates at which earnings are mapped
  into stock prices.
• Thus, one may expect CE to have higher ERCs than
  NCE.

39
Effect of Delayed Responses on Value Relevance of
CE NCE (Cont.)
40
Effect of Delayed Responses on Value Relevance of
CE NCE (Cont.)

• When only the contemporaneous association between
  returns and earnings are considered, ERCs on CE
  is not higher than those on NCE.
• The market does not value CE more than NCE.
• When the return window is extended to include
  year t1 so that correction for mispricing can be
  considered
• Coefficients on CE increase and become higher
  than those on NCE.
• Be cautious when inferring the value relevance of
  the accounting numbers of Chinese listed firms by
  contemporaneous returns-earnings association.

41
Conclusion

• Core earnings are more persistent than non-core
  earnings.
• But the market does not understand such
  difference.
• Core non-core earnings have similar ERCs.
• The implied persistency is lower than actual
  value for CE but higher than actual value for
  NCE.
• Profitable portfolios can be formed by the
  information contained in current earnings
  composition.
• The mispricing of ?CEt and ?NCEt are incremental
  to each other in the regression analysis.

42
Implications

• An out-of-sample analysis on the accumulated
  evidence on mispricing obtained in the U.S.
• Helpful to reassess how value-relevant are the
  financial data disclosed by Chinese firms.
• Direct implications for equity investors who are
  interested in Chinas capital market.
• Policy implications for Chinese regulators
  responsible for disclosure issues.

43

• THANK YOU!