how to estimate this equation?

[yuentk]

I want to examine the asymmetric relationship between the returns on the stock index and the returns on volatility by separating the positive and negative returns on volatility, the regression I want to estimate looks like this: R_Index C R_PV R_NV

Both sets of data are based on daily returns. I tried to create two series for positives and negatives, but i noticed there are some blanks between observations because if there is a positive return then the negative will be blank. When I tried to estimate using the ols, i got an error saying insufficient number of observations.
I did manage to regress with only one independent variable, either the r_pv or the r_nv, but i want to combine the two together. So how should I regress the equation above?

[startz]

You probably want zeros in place of the blank observations. How did you create your series?

[yuentk]

You probably want zeros in place of the blank observations. How did you create your series?

I created three series: for r_index, I entered daily returns observations from day 1 to day 1000; for r_pv, i entered changes of daily volatility(i.e. t - t(-1)), and the same for r_nv.
But as I mentioned before if let say on day 10 the change is a positive number, I will enter that number in r_pv on the 10th observation. This means there will be a "n/a" shown in the series of r_nv, vice versa.
I tried to estimate the equation with only one of the independent variables: r_index c r_pv or r_index c r_nv, both are working fine but the number of observations will be adjusted automatically. However, if i estimate both together, the error message pops up saying insufficient number of observations.

[startz]

You probably want zeros in place of the blank observations. How did you create your series?

I created three series: for r_index, I entered daily returns observations from day 1 to day 1000; for r_pv, i entered changes of daily volatility(i.e. t - t(-1)), and the same for r_nv.
But as I mentioned before if let say on day 10 the change is a positive number, I will enter that number in r_pv on the 10th observation. This means there will be a "n/a" shown in the series of r_nv, vice versa.
I tried to estimate the equation with only one of the independent variables: r_index c r_pv or r_index c r_nv, both are working fine but the number of observations will be adjusted automatically. However, if i estimate both together, the error message pops up saying insufficient number of observations.

Instead of NA in r_nv, you want a zero.