Dear all,
Currently i'am writing my master thesis about the idiosyncratic volatility puzzle. I need to replicate the paper of Hou and Loh (2016). This methodology consists of 3 stages:
1.First a Fama-MacBeth regression of the monthly returns on the lagged 1 month IVOL: π ππ‘ = πΌπ‘ + πΎπ‘πΌπππΏππ‘β1 + πππ‘. (I used: asreg month_RETW1 lag_ivol, fmb)
2. in the second stage they regress IVOL on the candidate variable: πΌπππΏππ‘β1 = ππ‘β1 + πΏπ‘β1πΆπππππππ‘πππ‘β1 + πππ‘β1. (now i used: by gvkey: asreg lag_ivol Candidate to extract the beta's)
3. In the last stage i need to decompose the regression coefficient from stage 1, where i need to use the regression coefficients πΏπ‘β1πΆπππππππ‘πππ‘β1 of stage 2 : πΎπ‘ = Cov[π ππ‘,πΌπππΏππ‘β1] / Var[πΌπππΏππ‘β1] = Cov[π ππ‘, (πΏπ‘β1πΆπππππππ‘πππ‘β1 + ππ‘β1 + πππ‘β1)] / Var[πΌπππΏππ‘β1] = Cov[π ππ‘, πΏπ‘β1πΆπππππππ‘πππ‘β1] / Var[πΌπππΏππ‘β1] + Cov[π ππ‘, (ππ‘β1 + πππ‘β1)] / Var[πΌπππΏππ‘β1] = πΎπ‘C + πΎπ‘R. From this i need to obtain standard errors of the fractions using the multivariate delta method.
I do not know exactly how i can extract the Beta's from stage 2, needed for the covariance term in stage 3. Also, i never heard of the multivariate delta method and Google does not (at least for me...) explain it properly.
Is there maybe anyone that has replicated this paper and may be able to help me?
Thanks in advance for the help
Yours sincerely,
Marijn Wolkers
Currently i'am writing my master thesis about the idiosyncratic volatility puzzle. I need to replicate the paper of Hou and Loh (2016). This methodology consists of 3 stages:
1.First a Fama-MacBeth regression of the monthly returns on the lagged 1 month IVOL: π ππ‘ = πΌπ‘ + πΎπ‘πΌπππΏππ‘β1 + πππ‘. (I used: asreg month_RETW1 lag_ivol, fmb)
2. in the second stage they regress IVOL on the candidate variable: πΌπππΏππ‘β1 = ππ‘β1 + πΏπ‘β1πΆπππππππ‘πππ‘β1 + πππ‘β1. (now i used: by gvkey: asreg lag_ivol Candidate to extract the beta's)
3. In the last stage i need to decompose the regression coefficient from stage 1, where i need to use the regression coefficients πΏπ‘β1πΆπππππππ‘πππ‘β1 of stage 2 : πΎπ‘ = Cov[π ππ‘,πΌπππΏππ‘β1] / Var[πΌπππΏππ‘β1] = Cov[π ππ‘, (πΏπ‘β1πΆπππππππ‘πππ‘β1 + ππ‘β1 + πππ‘β1)] / Var[πΌπππΏππ‘β1] = Cov[π ππ‘, πΏπ‘β1πΆπππππππ‘πππ‘β1] / Var[πΌπππΏππ‘β1] + Cov[π ππ‘, (ππ‘β1 + πππ‘β1)] / Var[πΌπππΏππ‘β1] = πΎπ‘C + πΎπ‘R. From this i need to obtain standard errors of the fractions using the multivariate delta method.
I do not know exactly how i can extract the Beta's from stage 2, needed for the covariance term in stage 3. Also, i never heard of the multivariate delta method and Google does not (at least for me...) explain it properly.
Is there maybe anyone that has replicated this paper and may be able to help me?
Thanks in advance for the help
Yours sincerely,
Marijn Wolkers