graduate rp

40
Analysis of the Relationship between Earnings from 10-Q and Stock Market Value Acc 692Q Accounting Research Project MAY 8, 2015 PACE UNIVERSITY Xun Zhou

Upload: maybel-xun-zhou

Post on 09-Feb-2017

76 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Graduate RP

Analysis of the Relationship between Earnings from 10-Q and Stock Market Value

Acc 692Q Accounting Research Project

MAY 8, 2015PACE UNIVERSITY

Xun Zhou

Page 2: Graduate RP

Analysis of the Relationship between Earnings

from 10-Q and Stock Market Value

Xun Zhou

Pace University

Abstract: During past decades, many researchers try to find one or more variable which can

predict the stock market. This paper intend to show the analysis between earnings from 10-Q and

stock market value. I try to find a possible relation during the analysis and find a significant

result. It is proved that Total receivable and Total Revenue act important role in the regression

model I built. This paper expect to provide a combined method of statistics and accounting, help

to provide solution for the situation in the market.

Purpose: To show analysis of the relation between the markets valuations of Earnings disclosed

on 10-Q by testing the relation between Market Value (stock price and stock returns) and

Earnings of financial reports 10-Q.

Design/methodology/approach: Use Descriptive Statistics test, Pearson Correlation and

Regression Model.

Key words: 10-Q Earnings; stock market valuation; person correlation; regression model

Data availability: data are available from sources identified in the paper.

Page 3: Graduate RP

1. Introduction

This paper intend to find out what may the current period earnings of company financial

reports affect the stock price and stock return on next period. It’s not a prediction, but a

potential relation between Market Value and Earnings disclosed. Cause we all know the

stock market is hard to predict, based on many reasonable or unreasonable factors. But this

paper try to show during the past 10 years : 2003 – 2012, how the stock market react to

financial reports (especially the 10-Q) disclosed, and may inspire the investors and

companies for further research. I try to combine with the recent hot issue “big data”, so in

this paper I use many statistics method like person correlation, regression model, residual

plot and scatter plot. I expect to provide a combined method of statistics and accounting, help

to provide solution for situation in the market.

There are many papers in the past did the research on or related to financial reports filings

and stock markets. Since it is obvious daily problem for investors and really profitable. In the

paper Information about Corporate Restructurings (see Bens, 2001), the author mentioned

that “several empirical studies have analyzed the association between performance and

disclosure finds that firms with negative earnings surprises are more likely to provide

earnings forecasts than those with good news.” In the paper Accruals Management and

Equity Valuation (see Balsam, Bartov and Marquardt, 2002), the authors gave the opinions

that investor sophistication plays a role in determining the timing of the market reaction to

Page 4: Graduate RP

accruals management, with the price reaction of sophisticated investors preceding that of

unsophisticated ones. The results suggest that investors reassess reported quarterly earnings

figures using other financial statement information and that this reassessment is associated

with substantial stock price change. Also in the paper Does the disclosure of corporate

governance structures affect firms’ earnings quality? (see Chang and Sun, 2010), “it is found

that the market valuation of earnings surprise is significantly higher for firms which disclose

stronger corporate governance functions....also that the effectiveness of corporate governance

in monitoring earnings management is improved after the mandated disclosure.” Based on

their findings, in this paper I chose the 10-Q for financial statements resource. The other

reason is that, during my data collection, I find out that 10K has too many year-end

adjustments, which the information affection are lagging for the stock market. What’s more,

10-Q are more fit for the frequently change stock market.

The reason use earnings as the variables is that most investors will put their eyes on earnings

first during making investment decision. Also generally companies are more willing to

disclose the earnings situation. Besides Total Revenues, Total Assets, Common Equity, Net

Income, and Income before Extraordinary Item: the traditional important information in

reports. Refer to the paper Market Valuation of Accrual Components (see Francis, 2008), the

author gave evidence support that market valuations for the receivable accrual are greater

than the valuations for other current accruals, and market valuations for cash flows are not

monotonically greater than the valuations for accruals. Based on his judgments, I also include

Total Receivable and Cash and Cash equivalent as variables. So there are total 8 variables

will be tested in this paper.

Page 5: Graduate RP

For the test objects, I choose Stock Price and Holding Period Returns (shows as Stock

Returns in this paper) which are the most investor most concerned factors. The Holding

Period Return is the total return on an asset or portfolio over a period during which it was

held. It is one of the simplest and most important measures of investment performance. It can

be calculated as:

HPR = (End Value - Initial Value) / Initial Value

where the End Value includes income, such as dividends, earned on the investment.

2. Theoretical Measurement

In order to control the test efficiency, I will first implement the Descriptive Statistics Table.

The table is used to show the range of my observations, since during the analysis, I hope the

data I used can sit in a comfortable range. The numbers too high or too low will affect the

test accuracy, so I will eliminate the unique data during sample selection. The descriptive

table will include all the elements, 8 variables from 10-Q and 2 stock variables from CRSP:

total 10 variables. Also will show the mean, median, maximize and minimize sample, which

will give a visualize process of the data.

For the next step, I choose the Person Correlation method. Correlation between sets of data is

a measure of how well they are related and the Pearson Correlation is the most common

measure of correlation in stats. The full name is the Pearson Product Moment Correlation or

Page 6: Graduate RP

PPMC. It shows the linear relationship between two sets of data. Two letters are used to

represent the Pearson correlation: Greek letter rho (ρ) for a population and the letter “r” for a

sample, the formula shows below:

The result for r will be between -1 and 1. High correlation: .5 to 1.0 or -0.5 to 1.0; Medium

correlation: .3 to .5 or -0.3 to .5; Low correlation: .1 to .3 or -0.1 to -0.3. But the limitation of

PPMC is that not able to tell the difference between dependent and independent variables. It

means it cannot prove which variable pull the trigger, but can only tell you whether there is a

relationship. So that’s why I will implement the regression model next to make the analysis more

reliable.

Regression analysis is a statistical process for estimating the relationships among variables. It

includes many techniques for modeling and analyzing several variables, when the focus is on the

relationship between a dependent variable and one or more independent variables. More

specifically, regression analysis helps one understand how the typical value of the dependent

variable (or 'criterion variable') changes when any one of the independent variables is varied,

while the other independent variables are held fixed. Most commonly, regression analysis

estimates the conditional expectation of the dependent variable given the independent variables –

that is, the average value of the dependent variable when the independent variables are fixed.

Regression models involve the following variables:

Page 7: Graduate RP

(1)The unknown parameters, denoted as β, which may represent a scalar or a vector.

(2)The independent variables, X.

(3)The dependent variable, Y.

So in this paper, the X will be 8 variables from 10-Q, which are the inputs in regression model,

and the Y will be the 2 categories from stock markets, which are outputs. I will separately set the

8 inputs for stock price and stock returns. So my regression model will be insert as:

SP t+1 = β0 + β1AR t + β2REV t + β3CE t + β4CF t + β5IBE t + β6TAn + β7NI t + β8WC t

+ ε t+1

SR t+1 = η0 + η 1AR t + η 2REV t + η 3CE t + η 4CF t + η 5IBE t + η 6TA t + η 7NI t + η

8WCn+ ε t+1

For here above: SP=Stock Price, SR=Stock Returns, AR=Total Receivable, REV=Total

Revenue, CE=Common Equity, CF=Cash and Cash Equivalent, IBE=Income before

Extraordinary Item, TA=Total Assets, NI=Net Income, WC=Working Capital.

Classical assumptions for regression analysis include:

The sample is representative of the population for the inference prediction.

The error is a random variable with a mean of zero conditional on the explanatory

variables.

The independent variables are measured with no error. (Note: If this is not so,

modeling may be done instead using errors-in-variables model techniques).

The predictors are linearly independent, i.e. it is not possible to express any

predictor as a linear combination of the others.

Page 8: Graduate RP

The errors are uncorrelated, that is, the variance–covariance matrix of the errors is

diagonal and each non-zero element is the variance of the error.

The variance of the error is constant across observations (homoscedasticity). If

not, weighted least squares or other methods might instead be used.

To provide the accuracy and effectiveness of the regression model, I will use the stepwise model

when implement the regression model. Stepwise regression includes regression models in which

the choice of predictive variables is carried out by an automatic procedure. Usually, this takes the

form of a sequence of F-tests or t-tests, but other techniques are possible, such as adjusted R-

square, Akaike information criterion, Bayesian information criterion, Mallows's Cp, PRESS, or

false discovery rate.

The frequent practice of fitting the final selected model followed by reporting estimates and

confidence intervals without adjusting them to take the model building process into account has

led to calls to stop using stepwise model building altogether or to at least make sure model

uncertainty is correctly reflected.

3. Sample Selection

The 10-Q sample are selected from Compustat (1), which is a database of financial,

statistical and market information on active and inactive global companies throughout the

world. This database provides a broad range of information products directed at

institutional investors, universities, bankers, advisors, analysts, and asset/portfolio

managers in corporate, M&A, private capital, equity, and fixed income markets. And the

Page 9: Graduate RP

stock market data are selected from CRSP (The Center for Research in Security Prices),

which is a provider of historical stock market data. For this paper, I chose from the North

American sector and monthly file for the stock market data. I pivot the monthly CRSP

data to quarterly first, and then joint it with Compustat by Tickers Symbol. The sample

must met the following criteria after joint:

(a) The company must have full 40 quarters (from October 2002 to September 2012)

records.

(b) The firm has a December fiscal year-end.

(c) The firm must have full records of the 8 variables including Total Assets, Common

Equity, Net Income, Total Receivable, Income before Extraordinary Item, Total

Revenue, Cash and Cash equivalent, and Working Capital.

(d) No consolidated filings.

(e) No firm with average stock price in the upper 10% or bottom 10%.

The reason I chose 40 quarters period from October 2002 to September 2012 is that the

affection to stock market is lagging to next period. Since the research period is from 2003

to 2012, the earning data need to be a quarter prior. In this paper, all the data are joint

lagging a quarter. I require a December fiscal year-end in (b) so that seasonal differences

across the calendar year are eliminated when I cross the data from earnings to stock

market. And the requirement for (d), no consolidated filings is considered that in CRSP,

parent company and subsidiary use different sticker symbols, which cannot be joint

appropriately with Compustat. So no consolidated filings firms can avoid this problem.

For criteria (e), too high or too low stock price will affect the accuracy during data

Page 10: Graduate RP

analysis, so I filter the firm with average stock price in the upper 10% range, or at the

bottom 10%.

After all the filters, the total observation company’s number is 769, each observation with

40 quarters data.

4. Hypothesis and Analysis

Since all the variables I choose are earnings, I make the hypothesis as:

H1: The Stock Price has positive relation with all the variables.

The first hypothesis are based on our common sense. Investors intend to buy the stocks

with high growth or potential growth. More earnings or more profit on the financial

reports naturally lead the investors willing to devote more on the stocks.

H2: The Stock Returns has negative relation with all the variables.

Since the Stock Returns based on (End Value - Initial Value) / Initial Value, higher the

original price, lower the Stock Return here. For example, company A’s initial price is 20

and company B’s initial price is 50, both of their price rise for 2 dollars in the same

period. Stock Return of company A equals to 0.1, but company B’s return is 0.04. So I

assume that company with strong price which may be more earnings, has a smaller

return. Furthermore, Balsam, Bartov and Marquardt had already prove that there is a

negative association between unexpected discretionary accruals and cumulative abnormal

returns around the 10-Q filing data in their paper.

Page 11: Graduate RP

H3: Total Receivable is the strongest variable that can affect Stock Price and Stock

Returns.

The idea choose Total Receivable as the leading variable is inspired by Francis I mention

above. He gave the idea that (2008,p.153) “Equity investors may find changes in account

payable to be less valuable that changes in accounts receivable due to the increased

persistence associated with revenue increases as compared to cost reductions.”

Furthermore, he also mentioned that(2008, p.153) “the good news associated with a

positive change in accounts receivable is more likely to exhibit future growth and

persistence than good news stemming from a cost-cutting strategy that is reflected in

accounts payable”. So I make the assumption that market valuation of the receivable is

greater than the market valuation for other earnings.

Table 1 is the Descriptive Statistics Test table for all 10 variables of 769 observations.

Observation period of Stock Price and Stock Returns are calendar year from 2003 to

2012. The other 8 variables from financial reports are from calendar year quarter 4 of

2002 to quarter 3 of 2012 due to the quarter lagging reflection need. Total 40 quarters

data.

Page 12: Graduate RP
Page 13: Graduate RP

Table 2 is the Pearson Correlation Matrix.

Page 14: Graduate RP

From table 2 we can see that:

All the variables are in the positive relation with Stock Price

All the variables are in the negative relation with Stock Returns.

The strongest 3 variables related to Stock Price are Total Revenue with R value 0.2023,

Income before Extraordinary Items with R value 0.1997, and Net Income with R value

0.1950.

The Strongest 3 variables related to Stock Returns are Total Assets with R value -0.0558,

Common Equity with R value -0.0533, and Total Receivables with R value -0.0527.

In the comprehensive observation, Receivables and Revenue are the two outstanding

variables with comparable strong relation to both Stock Price and Stock Returns.

Receivables has R value 0.1833 with Stock Price and -0.0527 with Stock Returns, and

Revenue has R value 0.2023 with Stock Price and -0.0462 with Stock Returns.

For further analysis, I insert the Liner Regression Model and implemented the stepwise

procedure. Stepwise regression includes regression models in which the choice of

predictive variables is carried out by an automatic procedure. It will automatically pick

out the suitable result, which means it can eliminate some no so reverent variables. So the

results of Stepwise Regression may not have total 8 variables.

Page 15: Graduate RP

Table 3 is the Regression Model of Stock Price after stepwise.

From table 3, we can see that the stepwise procedure automatically pick out the 7

variables. Among them, Common Equity, Income Before Extraordinary Items and

Revenue (2) these 3 variables have the smallest P value, which means these 3 variables

have the strongest relation during the total 8 variables this paper mention before. And the

Revenue and Income before Extraordinary Items are consistent with the Pearson

Correlation Matrix result.

Page 16: Graduate RP

Table 4 is the Regression Model of Stock Returns after stepwise.

The stepwise regression pick out only 5 significant variables in this model. Common

Equity, Net Income and Receivables (3) have the best performance. The results of

Common Equity and Receivables also consistent with the Correlation test.

To make sure the effective of the regression models, I insert an addition test, which is

the residual plot. Residual plot is a test assess the appropriateness of the model by

defining residuals and examining residual plots.

Page 17: Graduate RP

Here’s the Residual Plots Test for Stock Price Regression model:

Page 18: Graduate RP

Here’s the Residual Plots Test for Stock Returns Regression model:

Based on the Residual Plot test, I can make a conclusion that the Stock Price Regression

Model is more appropriate when dealing with the data sit in stock price range from 20 to

50, and the Stock Returns Regression Model is more appropriate when dealing with the

data sit in stock returns range from 120000 to 140000. Reflect to the Descriptive

Statistics table which is table 1, it shows that the Price Model will run better in the higher

half range while the Return Model will run better in the lower half range. It means, the

Price Model and the Return Model are suitable for the big companies. I assumed that the

result may cause for several reasons:

Page 19: Graduate RP

a. Large companies disclosed financial reports pattern are similar.

b. Most large companies with comparable long history and well-development stage,

and their stock market performance are tend to be more stable.

c. Most of large companies’ stock are holding by the fund, organization or financial

intermediate. Those shareholders are professional and sophisticated investor and

they have similar holding pattern.

Below are the graphic description of model results. I chose the Scatter Plot, which is a

type of mathematical diagram using Cartesian coordinates to display values for two

variables for a set of data. The data is displayed as a collection of points, each having the

value of one variable determining the position on the horizontal axis and the value of the

other variable determining the position on the vertical axis. A scatter plot is used when a

variable exists that is under the control of the experimenter. If a parameter exists that is

systematically incremented and/or decremented by the other, it is called the control

parameter or independent variable and is customarily plotted along the horizontal axis.

The measured or dependent variable is customarily plotted along the vertical axis. If no

dependent variable exists, either type of variable can be plotted on either axis or a scatter

plot will illustrate only the degree of correlation (not causation) between two variables.

(4)(5)

Page 20: Graduate RP
Page 21: Graduate RP
Page 22: Graduate RP
Page 23: Graduate RP
Page 24: Graduate RP

5. Conclusion

Based on the test and results. I can conclude that: for hypothesis 1 is true, earning variables

are positive related to Stock Price, and the 3 most relevant variables are Common Equity,

Income before Extraordinary Items, and Total Revenue. The hypothesis 2 is also true,

earning variables are negative related to Stock Returns, and the 3 most relevant variables are

Common Equity, Net Income and Total Receivables. For hypothesis 3, I prove that Total

Receivables indeed is one of the most useful variables during my choice in this paper.

Besides, Total Revenue and Common Equity can also be used to evaluate the Stock Price and

Stock Returns in future research. These two new found significant variables show good

performance in both Correlation test and Regression Model test.

It is proved in the model that Total Receivable and Total Revenue acting important role. This

is predictable since most investor will focus on these two factors when making decision,

besides, the new surprising findings from the model is that Common Equity is also a

significant factor. Common Equity shows comparable strong effect in both analyze with

Stock Price and Stock Returns. In further research dealing with stock market, researchers can

focus more on the Common Equity.

There are still some limitation of the paper. For example, I only include earnings as variables,

but in reality, accounts payable, long-term debt and other debts also effect on Stock Price and

Stock Returns a lot. But debt information are not as available as earnings since company

generally are willing to disclose the earnings. This paper shows quarterly change for stock

market value, which is still not detail enough, since stock market value change is a time

Page 25: Graduate RP

sensitive factor. Stock market can change sharply within a month, quarterly show may not so

appropriate fit the actual situation. Price Model will run better in the higher half range while

the Return Model will run better in the lower half range. For further research, the better way

may be input different range sector of price and returns, the results will be more accuracy.

I expect the analysis and method described in this paper provide a possible way when dealing

with this kind of situation in the future, and inspire more accounting issue combined with

“big data”.

Page 26: Graduate RP

Notes

1. All the number data from Compustat are in million (1000000)

2. The revenues mention in the paper is refer to Total Revenues.

3. The receivable mention in the paper is refer to Total Receivable.

4. The price mention in the paper is refer to Stock Price

5. The returns mention in the paper is refer to Stock Returns.

Page 27: Graduate RP

References

Balsam, Steven, Eli Bartov, and Carol Marquardt. "Accruals Management, Investor

Sophistication, and Equity Valuation: Evidence from 10-Q Filings." Journal of Accounting

Research 40.4 (2002): 987-1012. Web.

Bens, Daniel A. "The Determinants of the Amount of Information Disclosed about Corporate

Restructurings." Journal of Accounting Research 40.1 (2002): 1-20. Web.

Chang, Jui-Chin, and Huey-Lian Sun. "Does the Disclosure of Corporate Governance Structures

Affect Firms' Earnings Quality?" Review of Accounting and Finance 9.3 (2010): 212-43. Web.

CORREL function, Microsoft Office support pages, retrieved 2015-02-22

David A. Freedman, Statistical Models: Theory and Practice, Cambridge University Press

(2005)

Dechow, Patricia M., and Ilia D. Dichev. "The Quality of Accruals and Earnings: The Role of

Accrual Estimation Errors." The Accounting Review 77.S-1 (2002): 35-59. Web.

Doyle, Jeffrey T., Weili Ge, and Sarah Mcvay. "Accruals Quality and Internal Control over

Financial Reporting." The Accounting Review 82.5 (2007): 1141-170. Web.

Francis, Rick. "Market Valuation of Accrual Components." Review of Accounting and Finance

7.2 (2008): 150-66. Web.

Page 28: Graduate RP

Jarrell, Stephen B. (1994). Basic Statistics (Special pre-publication ed.). Dubuque, Iowa: Wm. C.

Brown Pub. p. 492. ISBN 0-697-21595-4.

R. Dennis Cook; Sanford Weisberg Criticism and Influence Analysis in Regression, Sociological

Methodology, Vol. 13. (1982), pp. 313–361

Rodney Ramcharan. Regressions: Why Are Economists Obessessed with Them? March 2006.

Accessed 2011-12-03.

Ronald A. Fisher (1954). Statistical Methods for Research Workers (Twelfth ed.). Edinburgh:

Oliver and Boyd. ISBN 0-05-002170-2.

Schmid Jr., John (December 1947). "The Relationship between the Coefficient of Correlation

and the Angle Included between Regression Lines". The Journal of Educational Research 41 (4).

JSTOR 27528906

Utts, Jessica M. Seeing Through Statistics 3rd Edition, Thomson Brooks/Cole, 2005, pp 166-

167. ISBN 0-534-39402-7