Beta Estimation
Exercises:
- Read in the clean CRSP data (
crsp_monthly) set fromcrsp_monthly.parquet(if you do not recall how to do this, check the previous exercise set) - Read in the Fama-French monthly market returns (
factors_ff_monthly) fromfactors_ff_monthly.parquet - Compute the market beta \(\beta_\text{AAPL}\) of ticker
AAPL(permno == 14593). You can use the functionlm()(R) orsmf.ols(Python) for that purpose (alternatively: compute the well-known OLS estimate \((X'X)^{-1}X'Y\) on your own). - For monthly data, it is common to compute \(\beta_i\) based on a rolling window of length five years. Implement a rolling procedure that estimates assets market beta each month based on the last 60 observations. You can use the package
slider(R),statsmodels.regression.rolling(Python), or a simpleforloop. (Note: this is going to be a time-consuming computational task) - Store the beta estimates as
beta_exercise.parquetindata-r/data-python(the filebeta.parquetalready contains the estimated values from the textbook - it may be a good idea to compare your results with the ones we get). - Provide summary statistics for the cross-section of estimated betas
- What is the theoretical prediction of CAPM concerning the relationship between market beta and expected returns? What would you expect if you create portfolios based on beta (you create a high- and a low-beta portfolio each month and track the performance over time)? How should the expected returns differ between high and low-beta portfolios?
Solutions: All solutions are provided in the book chapter Beta estimation (R version) or Beta estimation (Python version)