Quant lvl 2 - Reading 10

Question 1

Define multiple linear regression

Answer 1

tool that allows examination of the relationship (if any) between two types of variables

Question 2

Define partial slope coefficient (or partial regression coefficient)

Answer 2

slope coefficient in a multiple regression. “measures the expected change in dependent variable for 1unit increase in an independent variable, holding the other ind. variables constant.

Question 3

Define a dummy variable

Answer 3

qualitative independent variable in a regression. usually value of 1 to indicate true and value of 0 for false

Question 4

What are the three regression violations?

Answer 4

1. heteroskedasticity 2. serial correlation 3. multicollinearity

Question 5

Define heteroskedasticity

Answer 5

we assume that the variance of error is constant, but variance of error differs across observations

Question 6

What is the Breusch-Pagan test used for?

Answer 6

to test the heteroskedasticity. (Heteroskedasticity is “when the standard deviations of a variable, monitored over a specific amount of time, are non-constant” (investopedia))

Question 7

What is dummy variable regression

Answer 7

Tests whether the stock returns provide different average returns when the returns are related to other factors. e.g. takes on value of 1 if condition is true and 0 if it is false

Question 8

how many variables do you need for a dummy variable regression

Answer 8

n-1 since you want to distinguish among n categories.

Question 9

what is R-squared

Answer 9

coefficient of determination. measures “goodness of fit of estimated regression to the data”

Question 10

what is formula for R-squared (coefficient of determination)

Answer 10

total var-unexplained / total variation or the SSregression / SStotal

Question 11

define unconditional heteroskedasticity

Answer 11

“occurs when heteroskedasticity of error variance is not correlated with independ. variables in the multiple regression…creates no major problems for statistical inference”

Question 12

define conditional heteroskedasticity

Answer 12

heteroskedasticity in error variance that is correlated with the values of indep. variables in the regression.

Question 13

define multicollinearity

Answer 13

“high correlation in linear regression when the independent variables are highly correlated (2 or more)

Question 14

how do you correct multicollinearity?

Answer 14

omit one or more of teh correlated ind. variables

Question 15

3 way regression models are misspecified

Answer 15

1. functional form (important variables missing, data improperly pooled)
2. explanatory var. correlated with error term in time series model
3. other misspecifications resulting in nonstationarity