If you did not do well on the Level I CFA Program quant material then you may have to review a few of the concepts before tackling the level 2 material. The first reading, correlation and regression borrows heavily from the basics learned in the previous exam. While time series analysis is important, the first two readings seem to be the focus for the Institute so make sure you have a solid understanding.
The topic is not worth quite as many points as in the first exam but is still 5% – 10% of your total score and you will need some of the concept of correlation in other parts of the curriculum. The formulas can get a little mathematically intense but the concepts are pretty basic and understanding them will help get most of the points.
Correlation and Regression
Understand the simple linear relationship formula (Yi = b0 + b1X1 +e)
- Where Y is the dependent variable, the thing that you are trying to forecast and which is “dependent” on what happens with the other variable. X is the independent variable because it moves “independently” of the other variable in the equation (though there may be a deeper link between the two in reality, this is how I remember the difference). B0 is the intercept, the value of the dependent if X is zero. B1 is the slope and measures how much Y changes for movement in X. Know the assumptions behind the error term but don’t worry too much about it otherwise.
The Correlation Coefficient is extremely important, you need to understand the concept and be able to calculate it using the plot data or from the covariance. It ranges from -1 (perfect negative correlation) to +1 (perfect positive correlation) with 0 denoting no correlation between the variables.
- The approved calculators will take the plot data for X and Y and give you many of the needed outputs (averages, covariance, correlation and regression coefficients). You are supposed to know how to calculate these by hand…but the calculator can be used to check your answer.
- The Institute loves to test list items like limitations, strengths/weaknesses of a method and the situations in which a method is most appropriate. Make sure you understand the limitations to correlation analysis and regression.
- Assumes linear relationship, presence of outliers can distort, spurious correlation
You will need the material from the CFA level 1 to determine confidence intervals and significance, so you may want to review.
Multiple Regression can get tough but it is fairly important and you have a good chance of seeing a couple of questions. The formula and assumptions are basically the same as linear regression. Be able to read a computer output of the regression analysis and be able to plug the coefficients into the equation to estimate a dependent variable (i.e. if X1 and X2 are 15 and 10 then what is Y)
- Just multiply the coefficients (numbers in front of independent variables) by their respective variable then add/subtract the intercept term.
Heteroskedasticity, serial correlation and multicollinearity are very testable and you need to understand the definitions and tests you use to check them. Fairly easy stuff, just memorize the definition, the test and the effect on a regression.
- Heteroskedasticity – Breusch Pagan, incorrect standard errors
- Serial Correlation- Durbin Watson, incorrect standard errors and slope
- Multicollinearity- High R squared and low t scores on coefficients
Dummy variables are used to measure a yes/no state instead of a qualitative variable. Think seasonality, so an equation would have a dummy variable with a 1 if yes (maybe for a specific month) or a 0 for no (or all other months).
Time Series Analysis
The basic time series formula just uses time as the independent variable so is fairly easy to calculate. Serial correlation is an issue so understand how to tell if it is a problem and idea of seasonality.
- This is where the material gets more difficult. You should at least understand the reason behind the rest of the time series material (i.e. why you use autoregressive, mean reversion, random walk and residuals) but do not go crazy trying to become a mathematics guru. Spend a little more time on the seasonality portion but just get the basics for the rest.
Study session four starts the economics material for the CFA level 2 exam. Most of it is fairly conceptual but the currency exchange material can be a tough one for many candidates. Don’t fall behind and you should be able to work through it.
‘til next time, happy studyin’
Joseph Hogue, CFA