Hello all
My data is skewed at 0 because of my dependent variable, the dividend payout ratio, which is dividend amount/sales value (Dividends can be zero or a positive amount).
This ratio value is very small because the dividend is tiny compared to the sales.
Now I have two questions.
1. Many well-known researchers have used the Tobit model to explain the dividend model. But in this forum, I have read in many discussions that this kind of data is unsuitable for the Tobit model because the data hasn't any value below zero. That's why I am using panel data regression (FE Model). There is a heteroskedasticity issue, so I am running the Sargen-Hansen test (overidentification, xtoverid), which suggests using a fixed effect model. Is applying a fixed effect model right here? (Data is skewed at zero value)
2. The coefficients of the predictors are very small but significant (FE Model). What is the reason behind it, and is it not good to have a very small coefficient (.003,.004,.005, etc.)?
Thanks and Regards.
My data is skewed at 0 because of my dependent variable, the dividend payout ratio, which is dividend amount/sales value (Dividends can be zero or a positive amount).
This ratio value is very small because the dividend is tiny compared to the sales.
Now I have two questions.
1. Many well-known researchers have used the Tobit model to explain the dividend model. But in this forum, I have read in many discussions that this kind of data is unsuitable for the Tobit model because the data hasn't any value below zero. That's why I am using panel data regression (FE Model). There is a heteroskedasticity issue, so I am running the Sargen-Hansen test (overidentification, xtoverid), which suggests using a fixed effect model. Is applying a fixed effect model right here? (Data is skewed at zero value)
2. The coefficients of the predictors are very small but significant (FE Model). What is the reason behind it, and is it not good to have a very small coefficient (.003,.004,.005, etc.)?
Thanks and Regards.
Comment