Capital Markets Research (Using Archival Data) ØOriented towards financial accounting issues ØLinks with finance and economics

Questions? • Do earnings matter? (Ball and Brown, JAR, 1968; Beaver, JAR, 1968) • Are earnings components valued similarly? • Do earnings have incremental information content over cash flows • Does the perceived quality of an auditor affect the relation between earnings and returns? • Are disclosures about pension benefits or other postretirement benefits value relevant? • Would financial statements be more informative if GAAP were changed to permit managers to capitalize R&D?

Developments in Finance and Accounting Accounting Literature Positive Accounting Theory -- Watts and Zimmerman, 1986 • Predicts that the use of accounting numbers in contracting (compensation and debt contracts) and in the political process will influence a firm's reporting choices Valuation Theory -- Ohlson, CAR, 1995 • Provides accounting with theory, based on finance theory, on how earnings and book values map into prices.

Developments in Finance and Accounting Finance Literature Market efficiency -- Fama, 1970, 1971 -- prices fully reflect all available information CAPM -- Sharpe, 1964; Linter, 1965 -- a model of firm-specific expected returns

Relation between Accounting Equity Values & Stock Market Equity Values Accounting Measure of Wealth A0

-

L0

=

E0 + NI1 - Div 1

A1

-

L1

=

E1 + NI2 - Div 2

A2

-

L2

=

E2

Market’s Measure of Wealth

?

Price0

?

Return1 Price1

Common Market-Based Tests • Returns tests – Short window (Event study - News) – Long window (Association)

• Price tests

Short Window - Event Study • “Direct” link between information and use of information

• Event (loose definition) – Identifiable happening or announcement that you can pinpoint in time that provide NEW information relevant to the market

• Short window (days)

Event Study • Like --– Earnings announcement (Compustat, IBES) – Auditor change (press release, Dow Jones Interactive, Lexis/Nexis) – Change in accounting standard (press release, Dow Jones Interactive, Lexis/Nexis)

– New laws or regulations • Does the market react to this event (or information from this event)?

• Model: Unexpected Returns = Unexpected Earnings (and/or Other New Information)

Issues in Event Studies • • • • •

Event date? Model? Unexpected Returns = Unexpected Earnings Event window? (cumulative abnormal returns [CARs]) Expectation models? For returns – CAPM (estimate parameters in event window/non-event period and apply to event period) – market adjusted returns (adjust by market index or industry index or small cap and large cap index)

• For earnings (or other info) – – – –

Last year’s earnings Last year’s earnings with growth Time-series model Analysts forecasts

Long Window - Association • Link between information and use of that information • Longer window tests (year) • Two Basic Approaches --

Earnings Responses Coefficient Model? (Un)expected Returns = Unexpected Earnings (Un)expected Returns = INT + ERC*Unexpected Earnings + e • Earnings are like an annuity • So, an Earnings Responses Coefficient (ERC) should be near 1 + 1/rate (where r is the risk adjusted discount rate for equity) Economic Determinants of ERC-• persistence, risk, growth and interest rate • à control for in studies •

Kormendi and Lipe, JOB, 1987; Easton Zmijewski, JAE, 1989; Collins and Kothari, JAE, 1989

Some Problems with ERC Studies • • • •

heavily statistical based -- time series of earnings predictive ability do not control for accounting methods – ERC coefficients too low (1 to 3) - price lead earnings - inefficient capital markets - GAAP deficient - transitory earnings

Another Approach Model? Returns = Earnings + Changes in Earnings (Easton Harris, JAR, 1991) • Earnings are an increase in the book equity value and returns are an increase in the market equity value. • So the coefficient on earnings should be one. • Changes in earnings are like an annuity so coefficient should be near 1 + 1/rate Some similar problems -- price lead earnings - inefficient capital markets - GAAP deficient - transitory earnings

Price Tests -- the longest window! -- Ohlson theoretical work Model Variations: 1. Price = Earnings Earnings are an annuity So, the coefficient on earnings should be near 1 + 1/rate where r is the risk adjusted discount rate for equity 2. Price = Book Value Book value is the accountants' measure of equity Market Value is the Market's measure of equity So, the coefficient on book value should equal one.

Price Tests (continued) 3. Price = Book Value + Earnings Price is a weighted average of book values and earnings So, weights are also part of coefficients Barth, Beaver, Landsman, JAE, 1998 Collins, Maydew, Weiss, JAE, 1998 4. Price = Book Value+Residual Income (or Abnormal Earnings) where residual income=current year's earnings - beginning book value * r

Some Problems --

value of r

-

persistence of earnings

-

GAAP deficient (for 4 it should not matter)

-

GAAP conservative

-

transitory earnings

-

econometric issues -- size

Other Areas to Investigate • disclosure versus recognition • financial statement analysis • properties of accounting earnings related to corporate governance, ownership… • market inefficiency explanations versus behavioral explanations related to accounting anomalies • usefulness of accounting information in credit markets • international accounting issues

Useful Tools • One page summary of research idea, research question and motivation • Kinney paragraphs – What is the research question? – Why do we care? (motivation) – How are we going to answer?

• Libby boxes

– Economic construct – Empirical proxies (& explanation) – Variable description (& data source)

Relation between Accounting Equity Values & Stock Market Equity Values