FIN213

INTRODUCTORY STATISTICS
FOR
Week 9 FINANCIAL
Chapter 8ANALYSIS
– Multiple Linear Regression
 Using Dummy Variables
Financial analysts often need to use qualitative variables as independent variables in a regression.

A dummy variable is qualitative variable that takes on a value of 1 if a particular condition is true and 0 if
that
condition is false.
– used to account for qualitative variables such male or female, month of the year effects, etc.

 It may reflect an inherent property of the data (e.g., belonging to an industry or a region). For example, a
company belongs to health care industry (dummy variable = 1) or it does not (dummy variable = 0).

 It may be an identified characteristic of the data. We may introduce such a binary variable by a
condition that is either true or false. For example, the date may be before 2008 (prior to the onset of
financial crisis, dummy variable = 0) or after 2008 (after the onset of the financial crisis, dummy
variable = 1).

 Alternatively, it may be constructed from some characteristic of the data. The dummy variable would
reflect a condition that is either
Institute,
20200907.
true orInvestment
CFA. Quantitative false. Examples would
Analysis, 4th Edition. include
John satisfying
Wiley & Sons P&T, a condition, such as 2
 VIOLATIONS OF REGRESSION
ASSUMPTIONS
Inference based on an estimated regression model rests on certain assumptions being met.

Violations may cause the inferences made to be invalid.

So far, we have made an important assumption that the variance of error in a regression is constant
across observations (variance of errors are homoskedastic)

Heteroskedasticity occurs when the variance of the errors differs across observations.
 does not affect consistency
 causes the F test for the overall significance to be unreliable.
 t tests for the significance of individual regression coefficients are unreliable because
heteroskedasticity
introduces bias into estimators of the standard error of regression coefficients.

Institute, CFA. Quantitative Investment Analysis, 4th Edition. John Wiley & Sons P&T, 3
20200907.
 Homo vs Heteroskedastic variance
Regression with Regression with
homoskedasticity heteroskedasticity

Homoskedastic: There is no systematic relationship between the value of the independent variable and the regression residuals (the
vertical distance between a plotted point and the fitted regression line). Heteroskedastic: here, a systematic relationship is visually
apparent: On average, the regression residuals grow much larger as the size of the independent variable increases.

Institute, CFA. Quantitative Investment Analysis, 4th Edition. John Wiley & Sons P&T, 4
20200907.
 Types of Heteroskedasticity

• Unconditional Heteroskedasticity – This is heteroskedasticity of the

error term that is not correlated with the values of the independent
variable(s) in the regression. This form of heteroskedasticity does not
cause major problems for statistical inference.

• Conditional Heteroskedasticity – This is heteroskedasticity in the error

variance that is correlated with the values of the independent variable(s)
in the regression. This type causes the most problems for statistical
inference. Fortunately, many statistical software packages can easily test
and correct for conditional heteroskedasticity.

Institute, CFA. Quantitative Investment Analysis, 4th Edition. John Wiley & Sons P&T, 5
20200907.
 Testing for Heteroskedasticity

The Breusch–Pagan test consists of regressing the squared residuals from the
estimated regression equation on the independent variables in the regression.

–If no conditional heteroskedasticity exists, the independent variables will

not explain much of the variation in the squared residuals.
–If conditional heteroskedasticity is present in the original regression, however,
the independent variables will explain a significant portion of the variation in the
squared residuals.

Institute, CFA. Quantitative Investment Analysis, 4th Edition. John Wiley & Sons P&T, 6
20200907.
 Correcting for Heteroskedasticity

Two different methods to correct the effects of

conditional heteroskedasticity:

–computing robust standard errors, corrects the standard errors of

the linear regression model’s estimated coefficients to account for the
conditional heteroskedasticity.
–Generalized least squares, modifies the original equation in
an attempt to eliminate the heteroskedasticity.

Institute, CFA. Quantitative Investment Analysis, 4th Edition. John Wiley & Sons P&T, 7
20200907.
 Serial Correlation
When regression errors are correlated across observations, we say that they are serially
correlated (or autocorrelated)

• Serial correlation most typically arises in time series regressions.

• The principal problem caused by serial correlation in a linear regression is an incorrect

estimate of the regression coefficient standard errors

• Positive serial correlation is serial correlation in which a positive error for one
observation increases the chance of a positive error for another observation.
• Negative serial correlation, a positive error for one observation increases the chance of a
negative error for another observation, and a negative error for one observation increases
the chance of a positive error for another.

Institute, CFA. Quantitative Investment Analysis, 4th Edition. John Wiley & Sons P&T, 8
20200907.
 Testing for Serial Correlation
The Durbin Watson statistic is used to test for serial
correlation

• When the Durbin Watson (DW) statistic is less than 𝑑𝐿 , we reject the null hypothesis of no positive
serial correlation.

• When the DW statistic falls between 𝑑𝐿 and 𝑑 𝑢 , the test results are inconclusive

• When the DW statistic is greater than, 𝑑 𝑢 we fail to reject the null hypothesis of no positive
serial correlation

Institute, CFA. Quantitative Investment Analysis, 4th Edition. John Wiley & Sons P&T, 9
20200907.
 Durbin Watson statistic

Reject Do not Reject Ho

Inconclusive reject Ho, no Inconclusive
Ho Neg autocorr.
??? autocorr. ???
Positive
autocorr.
2
0 𝑑𝐿 𝑑𝑢 4-𝑑𝑢 4-𝑑𝐿 4

1. Obtain your DW critical values

2. Draw the number line
3. Find where the DWstat lies in the number
line
4. conclude

Institute, CFA. Quantitative Investment Analysis, 4th Edition. John Wiley & Sons P&T, 10
20200907.
 Correcting for Serial Correlation
Two alternative remedial steps when a regression has significant serial
correlation:

1. adjust the coefficient standard errors for the linear regression

parameter estimates to account for the serial correlation.

2. modify the regression equation itself to eliminate the serial

correlation.

Institute, CFA. Quantitative Investment Analysis, 4th Edition. John Wiley & Sons P&T, 11
20200907.
 Multicollinearity

Multicollinearity occurs when two or more independent variables (or

combinations
of independent variables) are highly (but not perfectly) correlated with each
other.

–does not affect the consistency of the OLS estimates of the regression
coefficients
–estimates become extremely imprecise and unreliable
•The classic symptom of multicollinearity is a high 𝑅2(and significant F-
statistic) even though the t-statistics on the estimated slope coefficients are not
significant.
•The most direct solution to multicollinearity
Institute, CFA. Quantitative is excluding
Investment Analysis, 4th Edition. John Wiley & Sons one
P&T, or more of the 12
20200907.
Problems in Linear Regression & their
Solutions

Institute, CFA. Quantitative Investment Analysis, 4th Edition. John Wiley & Sons P&T, 13
20200907.