We are trying to solve this exercise:
Compute for each Fama-French portfolio rolling monthly betas using a look-back window of 5 years, i.e., 60 months, such that you have the first betas by 31/12/1931 (or 01/01/1932). Specifically, starting on 31/12/1931 estimate the rolling betas for each of the 25 portfolios with a simple market model:
ri,t − rf,t = αi + βi(rm,t − rf,t) + i,t (1)
using the previous 60 monthly returns. Then, advance one month and estimate the 25 betas again using the previous 60 returns. Repeat this rolling beta estimation until 31/11/2019 (in Stata, rolling regressions by asset/identifier are facilitated with rangestat (ssc install required)).
Next, construct two aggregate portfolios based on the betas of the individual Fama-French portfolios. Specif- ically, at the beginning of each month, starting on 01/01/1932, assign the 5 Fama-French portfolios with the lowest betas (i.e., lowest quintile) to the low beta portfolio and the 5 Fama-French portfolios with the highest betas to the high beta portfolio. You can construct equally-weighted portfolios. Then, compute the monthly returns for the high and low beta portfolio up to 31/12/2019. As a third portfolio, construct the difference portfolio which is simply the return on the low beta portfolio minus the return on the high beta portfolio. Tabulate expected return, standard deviation, the Sharpe ratio, minimum return and maximum return for all three portfolios.
But we just get no Betas for January and February 1927 so rangestat doesn't get it that we want go back 59 months. Can you help me with the code or is there something with the date that doesn't work?
Compute for each Fama-French portfolio rolling monthly betas using a look-back window of 5 years, i.e., 60 months, such that you have the first betas by 31/12/1931 (or 01/01/1932). Specifically, starting on 31/12/1931 estimate the rolling betas for each of the 25 portfolios with a simple market model:
ri,t − rf,t = αi + βi(rm,t − rf,t) + i,t (1)
using the previous 60 monthly returns. Then, advance one month and estimate the 25 betas again using the previous 60 returns. Repeat this rolling beta estimation until 31/11/2019 (in Stata, rolling regressions by asset/identifier are facilitated with rangestat (ssc install required)).
Next, construct two aggregate portfolios based on the betas of the individual Fama-French portfolios. Specif- ically, at the beginning of each month, starting on 01/01/1932, assign the 5 Fama-French portfolios with the lowest betas (i.e., lowest quintile) to the low beta portfolio and the 5 Fama-French portfolios with the highest betas to the high beta portfolio. You can construct equally-weighted portfolios. Then, compute the monthly returns for the high and low beta portfolio up to 31/12/2019. As a third portfolio, construct the difference portfolio which is simply the return on the low beta portfolio minus the return on the high beta portfolio. Tabulate expected return, standard deviation, the Sharpe ratio, minimum return and maximum return for all three portfolios.
But we just get no Betas for January and February 1927 so rangestat doesn't get it that we want go back 59 months. Can you help me with the code or is there something with the date that doesn't work?
Comment