|
The Security Exchange Act requires the public companies announce the
sales of the previous month before the 10th the current month. It provides a
more timely accounting information for the investors. The monthly announce-
ment of sales numbers could be one of the reasons why prior researches on the
information content of earning announcements found no significant results.
Specifically, the following conditions should be of concern when researchers
try to explain the insignificant empirical evidence:
1. The monthly sales announcement contains information content. There-
fore the earnings announcement release only little incremental infor-
mation.
2. The investors should have utilized the more timely information to
forecast the earning. Yet, the earnings forecast models used by re-
searches did not consider the monthly sales information. It would
create a measurement errors on the unexpected earnings and consequent-
ly, result in a downward bias on the estimate of earning response
coefficients.
In order to provide for further evidence on the related issues, the ob-
jectives of this thesis are: 1. to examine the information content of unexpec-
ted sales; 2. to control for the measurement error of unexpected sales; 3. to
estimate the relationship of sales response coefficients and persistence of
unexpected sales; 4. to study the dynamic relation between stock prices and
monthly sales; 5. to compare the forecasting ability of modified earning fore-
casting model with that of the traditional models.
The results are as follow:
1. Sales response coefficient is positive and significant. Furthermore,
the SRC of good news samples is significantly smaller than that of bad
news samples.
2. To control for the measurement error of unexpected sales, the observa-
tions are grouped based on the CAR values. The grouped SRC is signi-
ficantly larger than the before-grouping SRC.
3. Based on the grouped observation, the SRC for persistent unexpected
sales is significantly larger than the SRC for non-persistent unex-
pected sales. If the sales process follows AR model, the correlation
coefficient for sales response coefficient and persistence response
coefficient is positive.
4. To explore the dynamic relation of stock price and sales, I establish
VARMA models for fabric industry, plastic industry, and electron in-
dustry separately. The result shows that sales numbers affect stock
prices, but not vise versa. It also implies the sales announcements
have information content.
5. The forecasting ability of the modified earnings forecast model is
better than the traditional models when forecasting gross profits and
earnings from operations. Using mean square error to measure fore-
casting ability, the modified earning forecast model is better on
forecasting net incomes before income taxes. Under the mean absolute
percentage error measure and comparing the performance on forecasting
net incomes before income taxes, the modified earning forecast model
is better than one of the traditional models and worse than the other
traditional model.
To sum up, this thesis shows that the monthly sales announcements contain
information content and the modified earning forecast model using monthly
sales information is better than traditional models. These results could help
researchers in explaining the insignificant empirical evidences on the infor-
mation content of the earnings announcements reported by prior studies.
|