Continuous disclosure and information asymmetry

Article information:

To cite this document: Mark Russell , (2015)," Continuous disclosure and information asymmetry ", Accounting Research Journal, Vol. 28 Iss 2 pp. 195 - 224 Permanent link t o t his document : http://dx.doi.org/10.1108/ARJ-11-2013-0085

Downloaded on: 25 March 2017, At : 04: 39 (PT) Ref erences: t his document cont ains ref erences t o 48 ot her document s. To copy t his document : permissions@emeraldinsight . com The f ullt ext of t his document has been downloaded 1816 t imes since 2015*

Users who downloaded this article also downloaded:

(2015),"The effect of financial crisis on auditor conservatism: US evidence", Accounting Research Journal, Vol. 28 Iss 2 pp. 160-171 http://dx.doi.org/10.1108/ARJ-06-2013-0033

(2015),"Earnings management, funding and diversification strategies of banks in Africa", Accounting Research Journal, Vol. 28 Iss 2 pp. 172-194 http://dx.doi.org/10.1108/ARJ-07-2013-0045

Access t o t his document was grant ed t hrough an Emerald subscript ion provided by emerald- srm: 602779 [ ]

For Authors

If you would like t o writ e f or t his, or any ot her Emerald publicat ion, t hen please use our Emerald

f or Aut hors service inf ormat ion about how t o choose which publicat ion t o writ e f or and submission guidelines are available f or all. Please visit www. emeraldinsight . com/ aut hors f or more inf ormat ion.

About Emerald www.emeraldinsight.com

Downloaded by UNIVERSITY OF INDONESIA At 04:39 25 March 2017 (PT)

Emerald is a global publisher linking research and pract ice t o t he benef it of societ y. The company

The current issue and full text archive of this journal is available on Emerald Insight at:

www.emeraldinsight.com/1030-9616.htm

Continuous disclosure and Disclosure

and

information asymmetry

information asymmetry

Mark Russell

UQ Business School, The University of Queensland, St. Lucia, Australia

Abstract Received 26 November 2013

Revised 31 May 2014 Purpose – This paper aims to examine whether firms with high information asymmetry disclose more

Accepted 15 August 2014 information under a continuous disclosure regime, and, second, the paper examines whether continuous disclosures reduce information asymmetry. Design/methodology/approach – The study models relations between continuous disclosures and information asymmetry using ordinary least squares regression and two-stage least squares regression. Findings – The study finds firms with high information asymmetry disclose more information. Further, the study finds that disclosure in the presence of high information asymmetry increases asymmetry. Finally, while bad news increases information asymmetry, the disclosure of firm-specific good and bad news is associated with reduced information asymmetry. Originality/value – The paper identifies conditions under which Continuous Disclosure Regime increases information in markets and influences information asymmetry.

Keywords Information asymmetry, Continuous disclosure, Firm performance expectations Paper type Research paper

1. Introduction

This paper evaluates whether firms with higher information asymmetry disclose more information under the Australian Continuous Disclosure Regime (CDR), and second, whether continuous disclosures reduce information asymmetry. Regulators require continuous disclosure from companies in an expanding number of countries and stock exchanges: Australia, Canada, New Zealand, Germany, New Zealand, Singapore, Hong Kong Stock Exchange, London Stock Exchange, New York Stock Exchange, American Stock Exchange and NASDAQ ( Oesterle, 1998 ; Table I ).

The study is motivated by the increased regulation of disclosure in markets to

Downloaded by UNIVERSITY OF INDONESIA At 04:39 25 March 2017 (PT)

alleviate information asymmetry. Information asymmetry leads to the inefficient alleviate information asymmetry. Information asymmetry leads to the inefficient

28,2 ARJ

I.

disclosure

across of

Monitoring and/or enforcement

Securities exchange

Disclosure principle Australia

Country agency

Statutory regulation

regulation

Australian Security and Investment

Continuous disclosure of price- Commission co-regulates with the

Corporations Act (Cth)

ASX Listing rules 3.1

sensitive information Australian Securities Exchange Canada

Canadian Securities

Continuous disclosure of material Administrators. Financial and

National Instrument

changes that would reasonably be Securities Commissions: British

51-102

expected to have a significant impact Columbia, Alberta, Saskatchewan,

on market price or value Manitoba, Ontario, Québec, New Brunswick, Nova Scotia

Securities Acts: (Ontario British Columbia, Alberta, Manitoba)

United Financial Services Authority

Continuous disclosure of information Kingdom

UK Listing Authority,

Listing Rule 9 (London

reasonably likely to have a

significant effect on price Germany

Stock Exchange)

Disclose any information that could Office

Federal Securities Supervisory

Securities Trading Act

1994, section 15

significantly influence the company’s stock price

Hong Kong Securities and Futures Commission

Securities and Futures

A listed corporation must, after any

Ordinance (Part XIVA)

inside information likely to have a material effect on price has come to its knowledge, disclose the information to the public

(continued)

Monitoring and/or enforcement

Securities exchange

Country agency

Disclosure principle New Zealand

Statutory regulation

regulation

Securities Commission of NZ

Securities Markets Act

NZX Listing rules 10.1.1

Continuous disclosure of price-

sensitive information Singapore

(NZ) 1988

Monetary Authority of Singapore

Securities and Futures

SGX Listing Rulebooks

Disclosure of information likely to

Act, section 203

Mainboard Rules Chapter 7

materially affect the price or value of

securities United States

Part 2

NYSE Listed Company

Release quickly to the public any

Manual section 202.05

news or information which might reasonably be expected to materially affect the market for its securities

NASDAQ Listing Rules

Disclosure of any material

information that would reasonably be expected to affect the value of their securities or influence investors’ decisions

asymmetry information Disclosure

Table

and

I.

ARJ

( Van Buskirk, 2012 ). Nevertheless, Australian CDR provides a data-rich environment to

reconcile conflicting disclosure studies and clarify the unknown effects of continuous disclosure.

The key contribution from this paper is to identify conditions under which CDR reduces information asymmetry and increases disclosure. First, the study provides evidence that firms with high information asymmetry disclose more information.

Second, the study indicates that disclosure in the presence of high information asymmetry increases asymmetry. Finally, while bad news is likely to increase information asymmetry, the disclosure of firm-specific good and bad news is associated with reduced information asymmetry.

The remainder of the paper is organized as follows. Section 2 reviews the Australian setting for continuous disclosure, the background literature and develops the hypotheses. Section 3 describes the research design, the empirical models and variables. The results are presented in Section 4, and Section 5 finishes the paper with conclusions.

2. Institutional background and hypothesis development

2.1 Continuous disclosure regulation The Australian CDR commenced in September 1994 following prominent corporate failures that involved delayed disclosure and a perceived lack of transparency in the Australian stock market ( Commonwealth Government Report of the Australian House of Representatives Standing Committee on Legal and Constitutional Affairs, 1991 ). The continuous disclosure rules are contained in ASX Listing Rule 3.1 as follows:

Once an entity is or becomes aware of any information concerning it that a reasonable person would expect to have a material effect on the price or value of the entity’s securities, the entity must immediately tell the ASX that information.

A reasonable person would be taken to expect information to have a material effect on the price or value of securities if the information would, or would be likely to, influence persons who commonly invest in securities in deciding whether or not to subscribe for, or buy or sell, the first mentioned securities (Section 677, Corporations Act 2001 (Cth)).

Listing Rule 3.1B provides exceptions to continuous disclosure which require confidentiality, reasonableness and one or more of the following conditions:

• a breach of law to disclose; • an incomplete proposal or negotiation;

Downloaded by UNIVERSITY OF INDONESIA At 04:39 25 March 2017 (PT)

• matters of supposition;

“exception” clauses have raised questions about the operation of the CDR ( Cassidy and

Disclosure

Chapple, 2003 ; Hsu, 2009 ). Applying the CDR requires management judgment on

and

price-sensitivity and price discovery in at least three dimensions. First, management must decide what information is price-sensitive and second whether information is

information

asymmetry

exempt from disclosure. Third, management must decide the content and presentation

of the information to be disclosed.

A number of studies consider the possibility that managers retain discretion in CDR

disclosure ( Brown et al., 1999 ; Hsu, 2009 ; Hsu et al., 2012 ). It is likely that CDR operates in line with earlier evidence that firms disclose by reference to their economic characteristics rather than strictly following disclosure rules ( Frost and Pownall, 1994 ).

Notwithstanding management incentives to selectively disclose, studies also suggest that CDR increased disclosure to change the Australian information environment. Hsu et al. (2012) study the properties of analyst forecasts under the CDR as it evolved from 1988 to 2001. Hsu et al. (2012) find evidence suggesting changes in analyst forecast properties, and after 1998, forecast dispersion deteriorated for small firms. Nevertheless, Corlett et al. (2000) suggest that CDR increased the disclosure of information. More specifically, Neagle and Tyskin (2001) examine regulatory activity and find small firms, loss firms and some industries are more likely to receive ASX price queries. Chan et al. (2007) and Dunstan et al. (2010) both find management earnings forecasts are associated with CDR.

The discretionary content and format of disclosure available to managers under CDR, and the conflicting incentives to disclose, potentially influence the impact of a mandatory disclosure regime on firms. Nevertheless, the study expects the economic benefits to the firm from increased disclosure and lower information asymmetry ( Jung and Kwon, 1988 ; Healy et al., 1999 ), and CDR compliance pressure, to give management incentives to disclose under CDR:

H1. Firms with higher information asymmetry disclose more information under CDR.

2.3 Implications of the continuous disclosure regime for information asymmetry The disclosure benefits for the firm from lower information asymmetry include the convergence of investor beliefs, stock liquidity increases and lower price volatility ( Lev, 1988 ). Nevertheless, higher day-to-day information asymmetry will persist for firms with more uncertain investments, longer-term projects and transactions

Downloaded by UNIVERSITY OF INDONESIA At 04:39 25 March 2017 (PT)

( Affleck-Graves et al., 2002 ; Miller, 2002 ). Furthermore, disclosure may increase

ARJ

that price-sensitive disclosures are associated with lower information based trading in

the post-2002 period. Consistent with the economic theory of disclosure consequences in markets ( Leuz and Verrecchia, 2000 ), the study expects CDR disclosure to reduce information asymmetry. Nevertheless, in line with event studies ( Krinsky and Lee, 1996 ), the study also expects CDR disclosure events to conditionally increase information asymmetry for

firms with high pre-existing information asymmetry. This leads to the following hypotheses:

H2. Price-sensitive disclosure under CDR reduces information asymmetry. H3. Price-sensitive disclosure under CDR for firms with high information

asymmetry increases information asymmetry.

3. Empirical analysis

3.1 Sample and data The initial sample includes all companies listed on the Australian Securities Exchange (ASX) between 1996 and 2006 with at least one year of market trading data. The sample for each test is determined by the model and measure of information asymmetry. Information asymmetry is measured annually and daily. The daily measurement of information asymmetry is affected by the large number of small firms in the ASX with thinly traded shares and zero trading days. To address these adverse trading effects, a daily sub-sample is based on the largest 500 ASX-listed firms over the period of 1996-2006, selected to exclude the firm-days with zero trading, zero price volatility and abnormally large bid-ask spreads (where a bid or ask has no corresponding ask or bid, respectively).

Price-sensitive disclosures come from the Securities Industry Research Centre of Asia-Pacific (SIRCA) signal G database. Financial and other company data and equity betas are from Aspect Huntley’s Fin Analysis database. Stock price and trading data for the ASX Securities Exchange Automated Trading System (SEATS) are from SIRCA databases. The bid-ask spreads for the ASX’s order-driven market come from SIRCA’s ASX Intra-day database, while the daily trading volume data and prices come from SIRCA’s ASX Daily database. Financial analyst data comes from I/B/E/S. The ASX’s industry classification is used until September 2002, after which the study uses the Standard and Poor’s Global Industry Classification System (GICS) as adopted by ASX in 2002.

Downloaded by UNIVERSITY OF INDONESIA At 04:39 25 March 2017 (PT)

3.2 Price-sensitive disclosure measure

(8) notice of meeting;

Disclosure

(9) stock exchange announcement;

and

(10) dividend announcement;

information

(11) progress report;

asymmetry

(12) company administration; (13) notice of call (Contributing shares);

(14) other; (15) chairman’s address; (16) letter to shareholders; (17) ASX query; (18) structured products; and (19) commitments test entity quarterly reports.

Similar to Brown et al. (1999) , the study measures price-sensitive disclosure (Disclosure i ) as the number of price-sensitive disclosures recorded by the ASX for firm i aggregated for day d and year t.

3.3 Information asymmetry measures Information asymmetry measures include the bid-ask spread, trading volume and price volatility, as used in the financial literature ( Leuz and Verrecchia, 2000 ).

Bid-Ask Spread – bid-ask spread is measured as the ratio of the quoted bid-ask spread to the quoted midpoint price for day d for firm i:

AskPrice i,d ⫺ BidPrice i,d

(AskPrice ⫹ BidPrice )

2 i,d

i,d

The measure is time-weighted where the weighting procedure is based on the number of seconds the quotation exists in each interval. Glosten and Milgrom (1985) model information asymmetry as a function of the bid-ask spread with a market maker. In markets such as the ASX that use electronic orders instead of a market maker,

Downloaded by UNIVERSITY OF INDONESIA At 04:39 25 March 2017 (PT)

information asymmetry is also captured by the bid-ask spread ( Brockman and Chung,

ARJ

traded for firm i (Ln(Volume)). A positive association is expected between trading

volume and price-sensitive disclosures.

3.4 Empirical models

3.4.1 H1 – continuous disclosure. Disclosure Model ( 1 ) is tested jointly with information

asymmetry Model ( 3 202 ) using a two-stage least squares regression model (2SLS):

Disclosure i,t ⫽ ␣ 0 ⫹ ␣ 1 Info_Asymmetry i,t ⫹ ␣ 2 Ln ( Size ) i,t ⫹ ␣ 4 Issue i,t ⫹ ␣ 5 Performance i,t ⫹ ␣ 6 Leverage i,t ⫹ ␣ 7 Analyst i,t

⫹ ␣ 8 Earn_Chg i,t ⫹ 兺 ␣ j Industry i,t ⫹␧ i

Where:

Disclosure i,t

⫽ the annual number of price-sensitive firm disclosures for firm i

for year t;

Info_Asymmetry i,t ⫽ the annual average of bid-ask spread, stock return volatility or annual aggregate of trading volume for firm i; bid-ask spread (Bid-ask spread) is measured as the ratio of the quoted bid-ask spread to the quoted midpoint price:

AskPrice i,d ⫺ BidPrice i,d

(AskPrice i,d ⫹ BidPrice i,d )

The measure is time-weighted, where the weighting procedure is based on the number of seconds the quotation exists in each interval, measured for firm i for day d, averaged annually;

⫽ trading volume (Ln(Volume)) equals the natural logarithm of the aggregate stock trading volume for firm i, measured for

year t; ⫽ stock price volatility (RET_Volatility) equals the standard

deviation of time-weighted trade to the trade price for firm i for

day d where standard deviation ⫽ (1/(n ⫺ 1) 冱[Price ⫺ Mean Price]2)1/2; n ⫽ the number of stock prices for firm i, mean

Downloaded by UNIVERSITY OF INDONESIA At 04:39 25 March 2017 (PT)

Analyst i,t ⫽ analysts’ following is an indicator variable taking the value of 1 if

Disclosure

sell-side analysts provide a public stock recommendation (buy,

and

hold or sell) for the stock at financial year-end t;

information

Earn_Chg

i,t

⫽ change in net profit after tax before abnormal items for year t scaled by average total assets for years t and t ⫺ 1 for firm i; and

asymmetry

Industry i,t ⫽ an indicator variable equal to 1 for an ASX industry group up to 2002 and GICS classification after 2002, and 0 otherwise for firm i

for year t.

The dependent variable in Model ( 1 ) is the financial-year aggregate of firm i price-sensitive disclosures. The independent variables in Model ( 1 ) include determinants of disclosure measured at financial-year end: firm size (Size) ( Lang and Lundholm, 1993 ); financing transactions proxied by Leverage and Issue ( Myers and Majluf, 1984 ); Analysts’ following (Analyst) ( Bhushan, 1989 ); performance (Performance) ( Lang and Lundholm, 1993 ); changes in performance (Earn_Chg); and industry ( Clinch and Sinclair, 1987 ). The variables earnings change and information asymmetry reflect the evidence that when perceived investor information asymmetry is high, a firms’ disclosure increases ( Lang and Lundholm, 1993 ).

3.4.2 H2 and H3 – information asymmetry and continuous disclosure parsimonious

model. The parsimonious information asymmetry Model ( 2 ) is tested using ordinary least squares (OLS) regression. In Model ( 2 ), information asymmetry is measured on day

d ⫹ 1 and price-sensitive disclosure is measured on day d. The study also runs Model ( 2 ) with information asymmetry measured on day d, to examine the speed of any

adjustment in the firm’s information asymmetry metrics, and found (untabulated) weaker but qualitatively similar results.

Info_Asymmetry i,d⫹1 ⫽ ␣ 0 ⫹ ␣ 1 Disclosure i,d ⫹ ␣ 2 Info_Asymmetry i,d⫺1 ⫹ ␣ 3 Disclosure i,d ⫻ Info_Asymmetry i,d⫺1

⫹ ␣ k Controls i,d ⫹␧ i

Where: Info_Asymmetry i,d ⫽ Information asymmetry is measured using three methods:

bid-ask spread, trading volume and stock return volatility;

Downloaded by UNIVERSITY OF INDONESIA At 04:39 25 March 2017 (PT)

daily bid-ask spread is measured as the ratio of the quoted bid-ask spread to the quoted midpoint price.

ARJ

Stock price volatility (RET_Volatility) equals the standard

deviation of time-weighted trade to the trade price for firm i for day d where standard deviation ⫽ (1/(n ⫺ 1) 冱[Price ⫺

Mean Price]2)1/2; n ⫽ the number of stock prices for firm i. The dependent variable in Model ( 2 ) is the bid-ask spread, trading volume or price

volatility, measured for firm i at day d ⫹ 1. MODEL ( 2 ) includes the number of price-sensitive disclosures (Disclosure) for day d, and an interaction term (Disclosure i,d ⫻

Info_Asymmetry i,d ⫺1). Model ( 2 ) also includes day d trading volume (Ln(Volume i,d )) and stock return volatility (RET_Volatility i,d ) in some estimations as control variables because of the correlations between the market-based variables ( Brown et al., 1999 ; Leuz and Verrecchia, 2000 ).

3.4.3 H2 and H3 – information asymmetry and continuous disclosure full model. Information asymmetry Models ( 3 ) and ( 4 ) are tested with OLS regressions. Second, to evaluate the possibility of bias from the non-random sampling of endogenous firm disclosure, the study tests Model ( 3 ) with Model ( 1 ) in a two-stage least squares regression.

Info_Asymmetry i,t ⫽ ␣ 0 ⫹ ␣ 1 Disclosure i,t ⫹ ␣ 2 Ln ( Size ) i,t ⫹ ␣ 3 Leverage i,t ⫹ ␣ 4 MBV i,t ⫹ ␣ 5 E_P i,t ⫹ ␣ 6 Sales_Growth i,t ⫹ ␣ 7 Asset_Growth i,t ⫹ ␣ 8 Earn_Chg i,t ⫹ ␣ 9 Accrual_Chg i,t

⫹ ␣ RET_Volatility ⫹ ␣ Ln ( Volume ) i,t (3)

10 i,t

⫹ ␣ 12 OwnerConc i,t ⫹ ␣ 13 Beta i,t ⫹ ␣ 14 RET_Mkt i,t ⫹␧ i

Info_Asymmetry i,t ⫽ ␣ 0 ⫹ ␣ 1 Disclosure i,t ⫹ ␣ 2 Ln ( Size ) i,t ⫹ ␣ 3 Leverage i,t ⫹ ␣ 4 MBV i,t ⫹ ␣ 5 E_P i,t ⫹ ␣ 6 Sales_Growth i,t ⫹ ␣ 7 Asset_Growth i,t ⫹ ␣ 8 Earn_Chg i,t ⫹ ␣ 9 Accrual_Chg i,t

⫹ ␣ 10 RET_Volatility i,t ⫹ ␣ 11 Ln ( Volume ) i,t (4) ⫹ ␣ 12 OwnerConc i,t ⫹ ␣ 13 Beta i,t ⫹ ␣ 14 RET_Mkt i,t

⫹ ␣ g Disclosure i,t ⫻ FirmExpectationProxies i,t ⫹␧ i

Downloaded by UNIVERSITY OF INDONESIA At 04:39 25 March 2017 (PT)

Where the additional variables are measured as:

Operations for firm i for year t, scaled by average total assets for

Disclosure

years t and t ⫺ 1;

and

OwnConc i,t ⫽ ownership concentration is the percentage of common shares on issue held by the top 20 shareholders measured at financial

information

asymmetry

year-end t for firm i;

Beta i,t ⫽ beta is the equity beta measured as the standard deviation of market adjusted share returns over 2 years for firm i for year t as

computed by Aspect Huntley; and

RET_Mkt i,d ⫽ market return is the annualized value-weighted average price return on all stocks in the SIRCA share price and price relatives dataset.

Information asymmetry is modeled as a function of the firm’s growth rate, performance and information environment ( Affleck-Graves et al., 2002 ; Brown et al., 2009 ). The information environment includes firm size, leverage and investment opportunities ( Lang and Lundholm, 1993 ; Leuz and Verrecchia, 2000 ). Following Penman (1996) and Tasker (1998) , investment opportunities and growth are measured using the market-to-book value of equity ratio (MBV), sales growth (Sales Growth) and asset growth (Asset Growth). Expectations of earnings growth are measured using the price-earnings ratio, inverted to earnings-price ratio (E_P) to avoid small or zero values in the denominator ( Penman, 1996 ). Earnings changes are included because earnings that change may be less informative than persistent earnings (Earn_Chg). Changes in accruals (Accruals_Chg) are defined as the change in the difference between earnings and operating cash flows scaled by total assets. The divergence between earnings and cash flow is linked to earnings management ( Lee et al., 1999 ).

Finally, studies suggest that firms with volatile earnings have disclosure incentives to reduce information asymmetry ( Miller, 2002 ). Model ( 4 ), therefore, includes interactions between disclosure and performance expectations measured by the market-to-book value of equity ratio, earnings-to-price ratio and changes in earnings and accruals as proxies for risk and growth (Disclosure ⫻ firm performance expectations).

Leuz and Verrecchia (2000) and Brown et al. (1999) find specific information environment associations between trading volume, firm size, stock return volatility, market index inclusion and ownership concentration. The study therefore includes stock beta (Beta), ownership concentration (OwnConc) and stock market return

(RET_Mkt) in some estimations of Models ( 3 ) and ( 4 ).

Downloaded by UNIVERSITY OF INDONESIA At 04:39 25 March 2017 (PT)

ARJ

28,2 Maximum

1.000 118.000 Bid-Ask Spread

0.048 75.879 Ln(Size)

⫺1.955 0.663 Earn Chg

⫺0.950 21.293 E_P

⫺1.355 0.313 Sales Growth

⫺1.493 1.575 Asset Growth

⫺0.963 1.328 Accrual Chg

⫺0.733 1.015 RET_Volatility

0.000 0.176 Ln(Volume)

⫺7.980 8.380 RET_Mkt

⫺0.316 0.391 Notes: Disclosure is the number of annual price-sensitive disclosures announced by firm i to the ASX

under the Continuous Disclosure regulations for firm i Bid-Ask Spread is the annual average bid-ask spread divided by the mid-point price computed by summing intraday quoted spreads; Ln(Size) is the natural logarithm of the market value of equity computed as ordinary shares on issue for firm i at financial year-end t multiplied by stock price at t; Issue is change in ordinary shares on issue divided by average ordinary shares for years t and t ⫺ 1; Leverage is total non-current liabilities for year t scaled by the average total assets for years t and t ⫺ 1; performance is net profit after tax before abnormal items divided by average total assets for years t and t ⫺ 1; Earn_Chg is the change in net profit after tax before abnormal items scaled by average total assets for firm i for years t and t ⫺ 1; MBV is the market value of firm i at t divided by the book value of equity for year t; E_P is reported earnings per share before abnormal items scaled by share price at t; Sales_Growth is the change in operating revenue scaled by the average total assets for firm i for year t; Asset_Growth is the change in total assets scaled by the average total assets for firm i for year t; Accrual_Chg is the change in income before tax, net interest, abnormal items minus cash flows from operations scaled by the average total assets for firm i for year t; RET_Volatility is the annual average standard deviation of the stock return; Log(Volume) is the natural logarithm of annual stock volume trade for firm i; OwnConc is the shares held by the top twenty shareholders divided by the total issued ordinary shares for year x; Beta is the monthly equity rate of

Downloaded by UNIVERSITY OF INDONESIA At 04:39 25 March 2017 (PT) return divided by the market return index measured as deviation from risk free rate for the previous 20

Table II.

months; RET_Mkt is the annualized value-weighted average price return on all stocks in the SIRCA

Disclosure and

Disclosure i,d

information

Bid-Ask Spread i,d

18.833 Ln(Volume) i,d

asymmetry

RET_Volatility i,d

5.295 Notes: Disclosure i,d is price-sensitive disclosure measured as the daily number of price-sensitive firm

disclosures announced by ASX for firm i for day d; Info_Asymmetry i,d is measured daily using three different variables: Bid-Ask Spread, trading volume and stock return volatility; daily bid ask spread (Bid-Ask Spread) as defined in Section 3.3 is measured as the ratio of the quoted bid-ask spread to the quoted midpoint price and then time-weighted, where the weighting procedure is based on the number of seconds the quotation exists in each interval, measured for firm i for day d; trading volume (Ln(Volume)) equals the natural logarithm of the daily share volume for firm i, measured for day d; stock price volatility (RET_Volatility) equals the standard deviation of the time-weighted trade to the trade price for firm i and day d where standard deviation ⫽ (1/(n ⫺ 1) 兺[Price – Mean Price]2)1/2 and n ⫽ the number of stock prices for firm i; samples are based on the largest 500 ASX-listed firms over the period

Table III.

1996-2006, selected to exclude the firm-days with zero trading, zero price volatility, and abnormally Descriptive statistics large bid-ask spreads (where a bid or ask has no corresponding ask or bid respectively). The descriptive

for the measures statistics of model ( 2 ) comprise approximately 859,432 firm-days; pooled data for the largest 500 ASX

used to estimate firms for 1996-2006, n ⫽ 859,432

model ( 2 ) bid-ask

Pearson correlations Spearman correlation

A B C D E F G H I A. Disclosure

0.038 B. Bid-Ask Spread

1 ⫺0.286 ⫺0.513 ⫺0.589 ⫺0.111 ⫺0.169 ⫺0.144 ⫺0.014 C. RET_Volatility

0.076 ⫺0.291 ⫺0.023 ⫺0.071 D. Ln(Volume)

0.168 ⫺0.009 E. Ln(Size)

0.044 F. Issue

0.090 ⫺0.006 G. Performance

0.336 H. Leverage

1 0.025 Downloaded by UNIVERSITY OF INDONESIA At 04:39 25 March 2017 (PT)

I. Earn Chg

ARJ

Consistent with the determinants of disclosure literature, disclosure is positively

correlated with firm size and leverage ( Myers and Majluf, 1984 ; Lang and Lundholm, 1993 ).

Table V reports correlations among the variables in the parsimonious Model ( 2 ). The information asymmetry measures for the bid-ask spread (bid-ask spread) for day

d, d ⫺ 1 and d ⫹ 1 are all negatively correlated with price-sensitive disclosures for day

d (Disclosure), while the trading volume (Ln(Volume)) and return volatility (RET_Volatility) measures are all positively correlated with price-sensitive disclosures for day d.

Table VI reports correlations among the variables in the information asymmetry

Model ( 3 ). The bid-ask spread is negatively correlated with price-sensitive disclosure, trading volume, size, leverage and earnings price ratio, consistent with prior evidence of higher information asymmetry for firms with lower disclosure ( Welker, 1995 ), lower trading volume ( Leuz and Verrecchia, 2000 ), smaller size ( Hasbrouck, 1991 ) and negative or ambiguous information about performance ( Brown et al., 2009 ; Ng et al., 2010 ). Trading volume (Ln(Volume)) is positively correlated with price-sensitive disclosure and size and leverage, consistent with lower information asymmetry for firms with more price-sensitive disclosure and greater firm size and leverage.

4. Results

4.1 Information asymmetry and price-sensitive disclosure 2SLS tests of Models ( 1 )

and ( 3 )

Table VII presents the 2SLS regression results of testing Models ( 1 ) and ( 3 )( Table VIII ). The disclosure Model ( 1 ) results give preliminary support for H1. Continuous disclosure is positively associated with bid-ask spread; hence, firms with higher information asymmetry are likely to increase disclosure ( Lang and Lundholm, 1993 ). The additional first-stage results show continuous disclosure is associated with trading volume but not with stock return volatility. Return volatility has many influences, and firms with high price volatility may not possess enough information to continuously disclose, as indicated by the low median of the disclosure variable in the descriptive

statistics. Untabulated OLS regression results of testing Model ( 1 ) are consistent with

Downloaded by UNIVERSITY OF INDONESIA At 04:39 25 March 2017 (PT)

2SLS results. The second-stage 2SLS results of using information asymmetry Model ( 2 ) are mixed

Pearson correlations Spearman correlation

A B C D E F G H I J A. Disclosure i,d

0.075 0.029 0.062 B. Bid-Ask Spread i,d

0.015 0.004 ⫺0.007 C. Bid-Ask Spread i,d⫺1

⫺0.010 0.015 ⫺0.013 D. Bid-Ask Spread i,d⫹1

0.004 ⫺0.007 0.019 E. Ln(Volume) i,d

0.362 0.280 0.280 F. Ln(Volume) i,d⫺1

0.283 0.361 0.269 G. Ln(Volume) i,d⫹1

1 0.280 0.271 0.361 H. RET_Volatility i,d

1 0.427 0.425 I. RET_Volatility i,d⫺1

0.578 1 0.391 J. RET_Volatility i,d⫹1

0.577 0.551 1 Notes: As indicated by the subscripts, information asymmetry measures in model ( 2 ) are computed on a daily basis and include measures for day d ⫹ 1,

day d ⫺ 1, and day d; Disclosure i,d is price-sensitive disclosure measured as the daily number of price-sensitive firm disclosures announced by ASX for firm i for day d; Info_Asymmetry i,d is measured daily using three different variables: Bid-Ask Spread, trading volume and stock return volatility; daily bid ask spread (Bid-Ask Spread) as defined in Section 3.3 is measured as the ratio of the quoted bid-ask spread to the quoted midpoint price and then time-weighted, where the weighting procedure is based on the number of seconds the quotation exists in each interval, measured for firm i for day d; trading volume (Ln(Volume)) equals the natural logarithm of the daily share volume for firm i, measured for day d; and stock price volatility (RET_Volatility) equals the standard deviation of the time-weighted trade to the trade price for firm i and day d where standard deviation ⫽ (1/(n ⫺ 1) 兺[Price ⫺ Mean Price]2)1/2 and n ⫽ the number of stock prices for firm i; samples are based on the largest 500 ASX-listed firms over the period 1996-2006, selected to exclude the firm-days with zero trading, zero price volatility, and abnormally large bid-ask spread (where a bid or ask has no corresponding ask or bid, respectively). The correlations in Table V comprise approximately 859,432 firm-days

estimate Correlations measures

asymmetry information Disclosure

model

Table

bid-ask used for

and

) to

( 2 the V.

3 ( estimate measures Correlations Table )-(

28,2 ARJ

4 ) models

VI.

used for to the

Pearson correlations Spearman correlation

L. M. N. O. A. Bid-Ask Spread

A B C D E F G H I J.

K.

0.002 0.307 ⫺0.160 ⫺0.086 B. RET_Volatility

⫺0.009 ⫺0.369 0.415 ⫺0.017 C. Ln(Volume)

⫺0.011 ⫺0.379 0.344 0.039 D. Disclosure

0.003 ⫺0.170 0.114 0.004 E. Ln(Size)

0.008 ⫺0.062 0.071 0.040 F. Leverage

0.038 0.104 ⫺0.077 ⫺0.043 G. MBV

0.020 ⫺0.060 0.119 0.107 H. E_P

0.120 0.084 ⫺0.141 0.019 I. Sales Growth

0.058 0.026 ⫺0.012 0.023 J. Asset Growth

0.132 ⫺0.007 0.006 0.051 K. Earns Chg

1 0.336 0.032 ⫺0.015 0.014 L. Accrual Chg

1 0.000 ⫺0.017 0.023 M. OwnConc

0.016 1 ⫺0.142 ⫺0.031 N. Beta

⫺0.020 ⫺0.148 1 ⫺0.013 O. RET_Mkt

0.036 ⫺0.033 0.007 1 Notes: Bid-Ask Spread is the annual average bid-ask spread divided by the mid-point price computed by summing intraday quoted spreads; RET_Volatility is the annual average standard

deviation of the stock return; Log(Volume) is the natural logarithm of annual stock volume trade for firm i; Disclosure is the number of annual price-sensitive disclosures announced by firm i to the ASX under the Continuous Disclosure regulations for firm I; Ln(Size) is the natural logarithm of the market value of equity computed as ordinary shares on issue for firm i at financial year-end t multiplied by stock price at t; Leverage is total non-current liabilities for year t scaled by the average total assets for years t and t ⫺ 1; performance is net profit after tax before abnormal items divided by average total assets for years t and t ⫺ 1; MBV is the market value of firm i at t divided by the book value of equity for year t; E_P is reported earnings per share before abnormal items scaled by share price at t; Sales_Growth is the change in operating revenue scaled by the average total assets for firm i for year t; Asset_Growth is the change in total assets scaled by the average total assets for firm i for year t; Earn_Chg is the change in net profit after tax before abnormal items scaled by average total assets for firm i for years t and t ⫺ 1; Accrual_Chg is the change in income before tax, net interest, abnormal items minus cash flows from operations scaled by the average total assets for firm i for year t; OwnConc is the shares held by the top twenty shareholders divided by the total issued ordinary shares for year x; Beta is the monthly equity rate of return divided by the market return index measured as deviation from risk free rate for the previous 20 months; RET_Mkt is the annualized value-weighted average price return on all stocks in the SIRCA share price and price relatives dataset; Sample of ASX firms based over the period 1996-2006 comprising approximately 5,249 firm-year observations

2SLS Estimates

Expected Ln(Volume) Expected Dependent variable

Info asymmetry Expected Bid-Ask Spread Expected

Bid-Ask

Expected RET_Volatility Expected

Disclosure sign Ln(Volume) Intercept

RET_Volatility

17.490*** Info asymmetry

⫺0.300 Ln(Size)

squares disclosure

asymmetry information

disclosure

Two-stage

Disclosure

asymmetry information estimates

Table

model least

and

and

VII.

of

Table

28,2 ARJ

VII.

2SLS Estimates

Info asymmetry Expected

Expected Dependent variable

Ln(Volume) sign Disclosure E_P

sign Bid-Ask Spread

RET_Volatility

⫺0.610 ⫺5.110*** Sales Growth

⫺0.030 0.270 Asset Growth

⫺0.540 ⫺1.880 Earns Chg

2.570 1.790 Accrual Chg

⫺0.120 ⫺1.050 RET_Volatility

0.990 44.990*** Ln(Volume)

RET_Mkt

Adjusted R 2 0.324

0.323 0.658 (continued)

2SLS Estimates

Info asymmetry Bid-Ask Spread

Ln(Volume) Dependent variable

RET_Volatility

Disclosure Ln(Volume) F-statistic

Disclosure

Bid-Ask Spread

Disclosure

RET_Volatility

4,594 4,594 Notes: Disclosure is the number of annual price-sensitive disclosures announced by firm i to the ASX under the Continuous Disclosure regulations for firm i Bid-Ask Spread is the annual average

bid-ask spread divided by the mid-point price computed by summing intraday quoted spreads; RET_Volatility is the annual average standard deviation of the stock return; Ln(Volume) is the natural logarithm of annual stock volume trade for firm i; Issue is change in ordinary shares on issue divided by average ordinary shares for years t and t ⫺ 1; Performance is net profit after tax before

abnormal items divided by average total assets for years t and t ⫺ 1; Analyst is an indicator variable taking the value of 1 if sell-side analysts provide a public stock recommendation (buy, hold or sell) for the stock at financial year-end t, and taking the value of 0 otherwise; Ln(Size) is the natural logarithm of the market value of equity computed as ordinary shares on issue for firm i at financial year-end t multiplied by stock price at t; Leverage is total non-current liabilities for year t scaled by the average total assets for years t and t ⫺ 1; MBV is the market value of firm i at t divided by the book value of equity for year t; E_P is reported earnings per share before abnormal items scaled by share price at t; Sales_Growth is the change in operating revenue scaled by the average total assets for firm i for year t; Asset_Growth is the change in total assets scaled by the average total assets for firm i for year t; Earn_Chg is the change in net profit after tax before abnormal items scaled by

average total assets for firm i for years t and t ⫺ 1; Accrual_Chg is the change in income before tax, net interest, abnormal items minus cash flows from operations scaled by the average total assets for firm i for year t; OwnConc is the shares held by the top twenty shareholders divided by the total issued ordinary shares for year x; Beta is the monthly equity rate of return divided by the market

return index measured as deviation from risk free rate for the previous 20 months; RET_Mkt is the annualized value-weighted average price return on all stocks in the SIRCA share price and price relatives dataset. Coefficient ␣ reported above t-statistic; *** significant at 1% level; ** significant at 5% level. Industry dummy variables and their 2SLS estimates are not reported; annual pooled sample of ASX-listed firms for the period 1996-2006

Disclosure i,t ⫽ ␣ 0 ⫹ ␣ 1 Info_Asymmetry i,t ⫹ ␣ 2 Ln ( Size ) i,t ⫹ ␣ 3 Issue i,t ⫹ ␣ 4 Performance i,t ⫹ ␣ 5 Leverage i,t ⫹ ␣ 6 Analyst i,t

⫹ ␣ 7 Earn_Chg i,t ⫹ ␣ 8 MBV i,t ⫹ ␣ 9 E_P i,t ⫹ ␣ 10 SalesGrowth i,t ⫹ ␣ 11 AssetGrowth i,t ⫹ ␣ 12 Accrual_Chg i,t

⫹ ␣ 13 OwnerConc i,t ⫹ ␣ 14 Ln(Volume) i,t ⫹ ␣ 15 RET_Volatility i,t ⫹ 兺 ␣ j Industry i,t ⫹␧ i

Information asymmetry model:

Info_Asymmetry i,t⫹1 ⫽ ␣ 0 ⫹ ␣ 1 Disclosure i,t ⫹ ␣ 2 Ln ( Size ) i,t ⫹ ␣ 3 Leverage i,t ⫹ ␣ 4 MBV i,t ⫹ ␣ 5 E_P i,t ⫹ ␣ 6 SalesGrowth i,t

⫹ ␣ 7 AssetGrowth i,t ⫹ ␣ 8 Earn_Chg i,t ⫹ ␣ 9 Accrual_Chg i,t ⫹ ␣ 10 RET_Volatility i,t ⫹ ␣ 11 Ln ( Volume ) i,t (3)

⫹ ␣ 12 OwnerConc i,t ⫹ ␣ 13 Beta i,t ⫹ ␣ 14 RET_Mkt i,t ⫹␧ i

asymmetry information Disclosure

Table

and

VII.

ARJ

28,2 Pr ⬎ t

GICS industry

2.050 0.0407 Capital_Goods

⫺0.330 0.739 Commercial_Professional_Services

Transportation

0.700 0.4867 Automobiles_Components

⫺1.030 0.3028 Consumer_Durables_Apparel

⫺1.160 0.247 Consumer_Services

⫺1.140 0.2523 Food_Staples_Retailing

⫺0.290 0.7721 Food_Beverage_Tobacco

⫺0.820 0.4101 Household_Personal_Products

0.120 0.908 Health_Care_Equipment_Services

Table VIII.

1.130 0.2584 GICS industry group Banks

Pharmaceuticals_Biotech_Sciences

⫺2.920 0.0035 variables and their

⫺0.490 0.6273 2SLS estimates

Diversified_Financials

⫺3.090 0.002 reported below for

Insurance

⫺1.120 0.2623 Disclosure model ( 1 )

Real_Estate

1.290 0.1971 with the Bid-Ask

Software_Services

1.070 0.2825 Spread measure of

Technology_Hardware_Equipment

⫺0.140 0.8848 information

Semiconductors_and_Equipment

0.220 0.8269 asymmetry

Telecommunication_Services

4.2 Information asymmetry and continuous disclosure tests from parsimonious

Model ( 2 )

Table IX presents results from parsimonious Model ( 2 ) which regress information asymmetry measures on disclosure, and on interactions between disclosure and day

d ⫺ 1 information asymmetry.

In columns one and two of Table IX , bid-ask spread and trading volume are negatively and positively associated with disclosure, respectively, consistent with H2. In column three, stock price volatility is not associated with disclosure. Table IX further

reports the interaction coefficients in Model ( 2 ). The study finds positive and significant

Downloaded by UNIVERSITY OF INDONESIA At 04:39 25 March 2017 (PT)

coefficient estimates for Disclosure d ⫻ Bid-AskSpread d⫺1 and Disclosure d ⫻

Disclosure

Information asymmetry measures

and

Variable

sign

Spread i,d⫹1

sign

Ln(Volume) i,d⫹1

sign

RET_Volatility i,d⫹1

information

asymmetry

Disclosure i,d

Bid-Ask Spread

i,d⫺1

Disclosure i,d ⫻ Bid-Ask Spread i,d⫺1

Ln(Volume) i,d⫺1

Disclosure i,d ⫻ Ln(Volume) i,d⫺1

RET_Volatility i,d⫺1

0.330 321.05*** Disclosure i,d ⫻ RET_Volatility i,d⫺1

0.163 39.05*** Ln(Volume) i,d

0.001 179.84*** RET_Volatility i,d

0.187 F-statistic

Adjusted R 2 0.615

Notes: Disclosure i,d is price-sensitive disclosure measured as the daily number of price-sensitive firm disclosures announced by ASX for firm i for day d; daily bid ask spread (Bid-Ask Spread) as defined in Section 3.3 is measured as the ratio of the

quoted bid-ask spread to the quoted midpoint price and then time-weighted, where the weighting procedure is based on the number of seconds the quotation exists in each interval, measured for firm i for day d; trading volume (Ln(Volume)) equals the natural logarithm of the daily share volume for firm i, measured for day d; and stock price volatility (RET_Volatility) equals the standard deviation of the time-weighted trade to the trade price for firm i and day d where standard deviation ⫽ (1/(n ⫺ 1) 兺[Price ⫺ Mean Price]2)1/2 and n ⫽ the number of stock prices for firm i. Coefficient ␣ reported above

t-statistic. *** significant at 1% level; samples for tests of model ( 2 ) comprise the largest 500 ASX-listed firms over the period 1996-2006, selected to exclude the firm-days with zero trading, zero price volatility and abnormally large bid-ask spreads

Table IX.

(where a bid or ask has no corresponding ask or bid, respectively) Information asymmetry and

Info_Asymmetry i,d⫹1 ⫽ ␣ 0 ⫹ ␣ 1 Disclosure i,d ⫹ ␣ 2 Info_Asymmetry i,d⫺1 ⫹ ␣ 3 Disclosure i,d

price-sensitive

disclosure tests from

Info_Asymmetry

parsimonious model Downloaded by UNIVERSITY OF INDONESIA At 04:39 25 March 2017 (PT)

i,d⫺1

␣ k Controls i,d ⫹␧ i

( 2 ) bid-ask

ARJ 28,2 Information asymmetry measures

Expected Variable

Expected

Bid-Ask Expected

RET_Volatility sign Ln(Volume) Intercept

4.890*** 14.420*** Ln(Size)

7.940*** 3.870*** E_P

⫺2.700*** ⫺11.410*** Sales Growth

1.320 ⫺0.950 Asset Growth

2.620*** ⫺0.530 Earns Chg

⫺3.450*** 2.640*** Accrual Chg

1.200 ⫺0.440 RET_Volatility

Ln(Volume)

0.002 12.440*** RET_Mkt

Disclosure ⫻ MBV

⫺3.030*** ⫺4.060*** Disclosure ⫻ E_P

⫺0.001 ⫺0.085 Downloaded by UNIVERSITY OF INDONESIA At 04:39 25 March 2017 (PT)

Table X.

Disclosure ⫻ EarnChg

Disclosure

Information asymmetry measures

and information

Variable

Bid-Ask Spread

RET_Volatility

Ln(Volume)

asymmetry

Notes: Bid-Ask Spread is the annual average bid-ask spread divided by the mid-point price computed by summing intraday quoted spreads; RET_Volatility is the annual average standard

deviation of the stock return; Ln(Volume) is the natural logarithm of annual stock volume trade for firm i; Disclosure is the number of annual price-sensitive disclosures announced by firm i to the ASX under the Continuous Disclosure regulations for firm i; Ln(Size) is the natural logarithm of the market value of equity computed as ordinary shares on issue for firm i at financial year-end t multiplied by stock price at t; Leverage is total non-current liabilities for year t scaled by the average total assets for years t and t ⫺ 1; MBV is the market value of firm i at t divided by the book value of equity for year t; E_P is reported earnings per share before abnormal items scaled by share price at t; Sales_Growth is the change in operating revenue scaled by the average total assets for firm i for year t; Asset_Growth is the change in total assets scaled by the average total assets for firm i for year t; Earn_Chg is the change in net profit after tax before abnormal items scaled by average total assets for firm i for years t and t ⫺ 1; Accrual_Chg is the change in income before tax, net interest, abnormal items minus cash flows from operations scaled by the average total assets for firm i for year t; OwnConc is the shares held by the top twenty shareholders divided by the total issued ordinary shares for year x; Beta is the monthly equity rate of return divided by the market return index measured as deviation from risk free rate for the previous 20 months; RET_Mkt is the annualized value-weighted average price return on all stocks in the SIRCA share price and price relatives dataset. Coefficient ␣ reported above t-statistic. *** significant at 1% level; ** significant at 5% level; annual pooled sample of ASX listed firms for the period 1996-2006

Info_Asymmetry i,t⫹1 ⫽ ␣ 0 ⫹ ␣ 1 Disclosure i,t ⫹ ␣ 2 Ln ( Size ) i,t ⫹ ␣ 3 Leverage i,t ⫹ ␣ 4 MBV i,t

⫹ ␣ 5 E_P i,t ⫹ ␣ 6 Sales_Growth i,t ⫹ ␣ 7 Asset_Growth i,t ⫹ ␣ 8 Earn_Chg i,t ⫹ ␣ 9 Accrual_Chg i,t ⫹ ␣ 10 RET_Volatility i,t ⫹ ␣ 11 Ln ( Volume ) i,t ⫹ ␣ 12 OwnerConc i,t ⫹ ␣ 13 Beta i,t ⫹ ␣ 14 RET_Mkt i,t

⫹ ␣ g Disclosure i,t ⫻ Firm Expections i,t ⫹␧ i

Table X.

Downloaded by UNIVERSITY OF INDONESIA At 04:39 25 March 2017 (PT)

In column 1 of Table XI , for the lowest two deciles of price volatility aggregated, In column 1 of Table XI , for the lowest two deciles of price volatility aggregated,

28,2 ARJ

for stock deciles

XI.

bid-ask for price of model

Highest decile lagged stock return volatility Variable

Lowest decile lagged stock return volatility

Expected sign 1 2 3 4 5 6 7 8 9 Intercept

53.600*** 49.150*** 66.870*** Disclosure i,d

⫺6.750*** ⫺3.860*** ⫺3.190*** Bid-Ask Spread i,d⫺1

272.010*** 281.680*** 294.170*** Disclosure i,d ⫻

7.700*** 3.930*** 2.680*** Spread i,d⫺1

Adjusted R 2 0.508

85,614 85,680 Notes: Disclosure i,d is price-sensitive disclosure measured as the daily number of price-sensitive firm disclosures announced by ASX for firm i for day d; daily bid-ask spread (Bid-Ask

Spread) as defined in Section 3.3 is measured as the ratio of the quoted bid-ask spread to the quoted midpoint price and then time-weighted, where the weighting procedure is based on the number of seconds the quotation exists in each interval, measured for firm i for day d; trading volume (Ln(Volume)) equals the natural logarithm of the daily share volume for firm i, measured for day d; and Stock price volatility (RET_Volatility) equals the standard deviation of the time-weighted trade to the trade price for firm i and day d where standard deviation ⫽ (1/(n ⫺ 1) 兺[Price – Mean Price]2)1/2 and n ⫽ the number of stock prices for firm i. Coefficient ␣ reported above t-statistic. *** significant at 1% level; ** significant at 5% level; the samples comprises the largest 500 ASX-listed firms over the period 1996-2006, selected to exclude the firm-days with zero trading, zero price volatility and abnormally large bid-ask spreads (where

a bid or ask has no corresponding ask or bid, respectively)

Info_Asymmetry i,d⫹1 (Bid-Ask Spread i,d⫹1 )⫽␣ 0 ⫹ ␣ 1 Disclosure i,d ⫹ ␣ 2 Info_Asymmetry i,d⫺1 (Bid-Ask Spread i,d⫺1 )

⫹ ␣ 3 Disclosure i,d ⫻ Info_Asymmetry i,d⫺1 (Bid-Ask Spread i,d⫺1 )⫹␧ i

4.3 Robustness tests

Disclosure

A number of tests are performed to check the robustness of the earlier results. The

and

parsimonious Model ( 2 ) is tested with all ASX stocks; Model ( 3 ) is regressed on changes in information asymmetry; and, in Model ( 2 ), clustered standard errors are

information

asymmetry

adjusted. These test results are discussed below, although untabulated due to space

constraints. The study runs parsimonious Model ( 2 ) using the full sample of all ASX-listed

companies for 1996-2006, with variables winsorized at 1 and 99 per cent percentiles to remove outliers. The Model ( 2 ) results for all-firm sample differ in two ways from the reported results in Table IX for the largest 500 ASX firms. The study finds the coefficient for price-sensitive disclosure changes from a negative sign (as in Table IX ) to

a positive sign, reflecting the inclusion of firms with relatively higher level of information asymmetry. The study also finds the sign of the Disclosure d ⫻ Bid-Ask Spread d⫺1 interaction changes from positive (as in Table IX ) to negative. This evidence confirms earlier results that the pre-existing level of information asymmetry is an important factor in tests of disclosure effects under CDR.

Second, Model ( 3 ) is run with a change of information asymmetry measure, rather than the level of information of asymmetry, as the dependent variable. The results show that the change of information asymmetry models has a lower adjusted R-square than the levels of asymmetry models. Nevertheless, for changes of bid-ask