Seminar Financial Accounting Research
Session 1
An empirical evaluation of accounting income numbers (Ball & Brown,
1968)
Introduction and motivation
Motivation:
- Accounting research was mainly normative, prescriptive or theoretical → shortcoming = it
ignored the extent to which predictions of the model conform to observed behaviour (i.e.
research was focused on “ideal world”, normative). For this reason, the authors
operationalise the usefulness of earnings numbers by looking at the decision-making
process of investors, which is reflected in security prices.
- Moreover, the prevailing view was that earnings numbers were useless; earnings are
historical data and were not considered important for the future. In addition, stock markets
were viewed as chaotic and irrational.
Coincidence in other research area → finance: framing stock price behaviour over time as a
function of information flows to the market. Therefore, there is a predictable pattern
around information events. Giving this pattern, you can use event studies as natural way of
studying price behaviour as a function of information.
RQ = Does net income have an impact on the investment decision, i.e., is it reflected in
security prices?
Idea → assess the usefulness of existing accounting numbers by examining their
information content and timeliness:
* Do earnings numbers capture information that is contained in stock prices? → do earnings
have an effect on the decision-making process of investors?
* Do earnings convey new information to the stock market, i.e., is it a timely medium to
investors?
Methodology
Event study → observe the market reaction around the earnings announcement (EA) date.
Event window → 12 months before and 6 months after the earnings announcement date:
* (-12:0) because of annual earnings (= information content)
* (0:+6) to give plenty of time to react to earnings (= timeliness)
Assumptions are made that investors:
- think EPS are relevant for decision making
- are able to forecast EPS through accounting information
- adjust their forecasting errors by changing the share price (i.e. markets are efficient)
Step 1: Unexpected earnings (= independent variable)
Market model → actual income (eps) – expected income (eps) = abnormal income (eps):
∆ I j , t−τ =^α 1 jt + α^ 2 jt ∆ M j , t−τ + u^ j , t−τ
- M is a market index of income
- Use OLS-method to calculate α^ 1 jt & α^ 2 jt
1
,- Get the expected income change this year → ∆ I^ jt =α^ 1 jt + α^ 2 jt ∆ M j ,t
- Realized income this year → ∆ I jt
- Calculate unexpected income (= forecast error) → u^ j ,t −τ (= ∆ I jt – ∆ I^ jt )
* u^ j ,t −τ > 0 → good news (GN) firm
* u^ j ,t −τ < 0 → bad news (BN) firm
Three approaches:
- Change in EPS (uncorrected; naïve model → ∆ I jt is entirely unexpected)
- Abnormal income
- Abnormal EPS
Step 2: Abnormal stock returns (= dependent variable)
1. Monthly stock return → PR jm=( p j ,m+ 1+ d jm)/ p jm
* p j ,m+ 1 = closing price
* d jm= dividend this month
* p jm = opening price
2. Decomposition of returns → expected (market) & unexpected part
[PR¿ ¿ jm−1]=b^ 1 j+ b^ 2 j [ Lm−1]+ v^ jm ¿
* Lm = market index
* PR jm = monthly price relative for firm j in month m
* ^v jm = stock return residual for firm j in month m (= abnormal stock return)
* Use OLS-method to estimate ^v jm separately for each firm-month
Step 3: Abnormal performance index (API)
Calculation steps of API:
1. Start with abnormal stock returns ^v jm (= residual)
2. Calculate cumulative abnormal stock returns for each firm-month (e.g. for month +6 and
firm j: cumulate abnormal stock return of firm j from m = -11 before to +6 months after EA)
3. Calculate average cumulative abnormal stock returns for each month across firms in 1)
GN sample, 2) BN sample and 3) full sample.
Step 4: Study the development of the API in the course of time around the announcement
date of the earnings information, separately for the good news (GN) sample and the bad
news (BN) sample
Methodology summarized, for each firm:
1. Estimate investors’ earnings expectations
* Classify each firm as GN (actual earnings > expected earnings) or BN
2. Estimate abnormal stock return for each firm-month
4. Calculate abnormal performance index (API)
* Start with month of EA (= month 0) and corresponding abnormal stock return for
each firm
* Cumulate abnormal stock return for each firm from -11 to 0
* Calculate average cumulative abnormal stock return for month 0 for each of the
samples
* Repeat for the months -1 to -12 and months +1 to +6
Results
2
, Main finding = firms with positive (negative) abnormal/unexpected earnings have positive
(negative) abnormal stock returns → earnings are useful for investors; the stock market
reacts to earnings announcements.
Three key findings:
- Information content of earnings
Of all information about an individual firm which becomes available during a year,
over 50% is captured in that year’s income number
- Prices lead earnings
Market anticipates earnings information, a large fraction (85-90%) is impounded
into prices prior the EA. Therefore, earnings are not timely.
- Prices lag earnings
There is a post-earnings announcement drift (i.e. markets are not completely
efficient) → significant cumulative abnormal returns also occur in the post-
announcement period and indicate that the market underreacts to earnings.
How Do Earnings Numbers Relate to Stock Returns? A Review of Classic
Accounting Research with Updated Evidence (Nichols & Wahlen, 2004)
Introduction and motivation
RQ = How do earnings numbers relate to stock returns?
Bottom-line measures for firm performance: accounting → earnings & capital market →
stock returns. This paper summarizes the theory and evidence on how accounting earnings
information relates to stock returns. In addition, they present new empirical evidence on
the relation between earnings and returns by replicating and extending three studies:
1. the association between earnings and returns
2. the effects of earnings persistence on the earnings-returns association
3. the quickness and completeness of the market’s reaction to earnings news
Theory
The three links/assumptions relating earnings to stock returns:
1. Current earnings numbers provide information that equity shareholders can use to form
expectations for future earnings
Earnings persistence refers to the likelihood that a firm’s earnings level will recur in
future periods (= essential element of link 1). In the case of transitory earnings, the
persistence is low or even zero (e.g. one-time gain/loss). Persistent earnings
contribute much more to share value than do transitory earnings.
2. Current and expected future profitability provides shareholders with information about
the firm’s expected future dividend-paying capacity (created wealth will ultimately be
distributed to shareholders through dividends)
3. Share price reflects the present value of all expected future dividends
These links imply that new accounting earnings information that triggers a change in
investors’ expectations about future dividends should correspond with a change in the
market value of the firm.
Methodology and results
3
Session 1
An empirical evaluation of accounting income numbers (Ball & Brown,
1968)
Introduction and motivation
Motivation:
- Accounting research was mainly normative, prescriptive or theoretical → shortcoming = it
ignored the extent to which predictions of the model conform to observed behaviour (i.e.
research was focused on “ideal world”, normative). For this reason, the authors
operationalise the usefulness of earnings numbers by looking at the decision-making
process of investors, which is reflected in security prices.
- Moreover, the prevailing view was that earnings numbers were useless; earnings are
historical data and were not considered important for the future. In addition, stock markets
were viewed as chaotic and irrational.
Coincidence in other research area → finance: framing stock price behaviour over time as a
function of information flows to the market. Therefore, there is a predictable pattern
around information events. Giving this pattern, you can use event studies as natural way of
studying price behaviour as a function of information.
RQ = Does net income have an impact on the investment decision, i.e., is it reflected in
security prices?
Idea → assess the usefulness of existing accounting numbers by examining their
information content and timeliness:
* Do earnings numbers capture information that is contained in stock prices? → do earnings
have an effect on the decision-making process of investors?
* Do earnings convey new information to the stock market, i.e., is it a timely medium to
investors?
Methodology
Event study → observe the market reaction around the earnings announcement (EA) date.
Event window → 12 months before and 6 months after the earnings announcement date:
* (-12:0) because of annual earnings (= information content)
* (0:+6) to give plenty of time to react to earnings (= timeliness)
Assumptions are made that investors:
- think EPS are relevant for decision making
- are able to forecast EPS through accounting information
- adjust their forecasting errors by changing the share price (i.e. markets are efficient)
Step 1: Unexpected earnings (= independent variable)
Market model → actual income (eps) – expected income (eps) = abnormal income (eps):
∆ I j , t−τ =^α 1 jt + α^ 2 jt ∆ M j , t−τ + u^ j , t−τ
- M is a market index of income
- Use OLS-method to calculate α^ 1 jt & α^ 2 jt
1
,- Get the expected income change this year → ∆ I^ jt =α^ 1 jt + α^ 2 jt ∆ M j ,t
- Realized income this year → ∆ I jt
- Calculate unexpected income (= forecast error) → u^ j ,t −τ (= ∆ I jt – ∆ I^ jt )
* u^ j ,t −τ > 0 → good news (GN) firm
* u^ j ,t −τ < 0 → bad news (BN) firm
Three approaches:
- Change in EPS (uncorrected; naïve model → ∆ I jt is entirely unexpected)
- Abnormal income
- Abnormal EPS
Step 2: Abnormal stock returns (= dependent variable)
1. Monthly stock return → PR jm=( p j ,m+ 1+ d jm)/ p jm
* p j ,m+ 1 = closing price
* d jm= dividend this month
* p jm = opening price
2. Decomposition of returns → expected (market) & unexpected part
[PR¿ ¿ jm−1]=b^ 1 j+ b^ 2 j [ Lm−1]+ v^ jm ¿
* Lm = market index
* PR jm = monthly price relative for firm j in month m
* ^v jm = stock return residual for firm j in month m (= abnormal stock return)
* Use OLS-method to estimate ^v jm separately for each firm-month
Step 3: Abnormal performance index (API)
Calculation steps of API:
1. Start with abnormal stock returns ^v jm (= residual)
2. Calculate cumulative abnormal stock returns for each firm-month (e.g. for month +6 and
firm j: cumulate abnormal stock return of firm j from m = -11 before to +6 months after EA)
3. Calculate average cumulative abnormal stock returns for each month across firms in 1)
GN sample, 2) BN sample and 3) full sample.
Step 4: Study the development of the API in the course of time around the announcement
date of the earnings information, separately for the good news (GN) sample and the bad
news (BN) sample
Methodology summarized, for each firm:
1. Estimate investors’ earnings expectations
* Classify each firm as GN (actual earnings > expected earnings) or BN
2. Estimate abnormal stock return for each firm-month
4. Calculate abnormal performance index (API)
* Start with month of EA (= month 0) and corresponding abnormal stock return for
each firm
* Cumulate abnormal stock return for each firm from -11 to 0
* Calculate average cumulative abnormal stock return for month 0 for each of the
samples
* Repeat for the months -1 to -12 and months +1 to +6
Results
2
, Main finding = firms with positive (negative) abnormal/unexpected earnings have positive
(negative) abnormal stock returns → earnings are useful for investors; the stock market
reacts to earnings announcements.
Three key findings:
- Information content of earnings
Of all information about an individual firm which becomes available during a year,
over 50% is captured in that year’s income number
- Prices lead earnings
Market anticipates earnings information, a large fraction (85-90%) is impounded
into prices prior the EA. Therefore, earnings are not timely.
- Prices lag earnings
There is a post-earnings announcement drift (i.e. markets are not completely
efficient) → significant cumulative abnormal returns also occur in the post-
announcement period and indicate that the market underreacts to earnings.
How Do Earnings Numbers Relate to Stock Returns? A Review of Classic
Accounting Research with Updated Evidence (Nichols & Wahlen, 2004)
Introduction and motivation
RQ = How do earnings numbers relate to stock returns?
Bottom-line measures for firm performance: accounting → earnings & capital market →
stock returns. This paper summarizes the theory and evidence on how accounting earnings
information relates to stock returns. In addition, they present new empirical evidence on
the relation between earnings and returns by replicating and extending three studies:
1. the association between earnings and returns
2. the effects of earnings persistence on the earnings-returns association
3. the quickness and completeness of the market’s reaction to earnings news
Theory
The three links/assumptions relating earnings to stock returns:
1. Current earnings numbers provide information that equity shareholders can use to form
expectations for future earnings
Earnings persistence refers to the likelihood that a firm’s earnings level will recur in
future periods (= essential element of link 1). In the case of transitory earnings, the
persistence is low or even zero (e.g. one-time gain/loss). Persistent earnings
contribute much more to share value than do transitory earnings.
2. Current and expected future profitability provides shareholders with information about
the firm’s expected future dividend-paying capacity (created wealth will ultimately be
distributed to shareholders through dividends)
3. Share price reflects the present value of all expected future dividends
These links imply that new accounting earnings information that triggers a change in
investors’ expectations about future dividends should correspond with a change in the
market value of the firm.
Methodology and results
3
