Create Cofficient Plot

[Maxence Morlet]

Hi all,

Apologies I do not have a data extract for this.

What code could I write to generate the picture attached?

The Y-axis represents the magnitude of each coefficient. The X-axis is a different lag of the same regressor.

What would also be fine is for each lag, get a point estimate with 95% confidence bands, and link all of these with a curve which passes through the point estimate at each lag.

Thanks!

[Andrew Musau]

I don't think the authors plotted the raw coefficients. It seems like some sort of smoothing has been applied. Here is some way, using a quadratic prediction plot.

webuse sp500, clear
tsset date
frame create results lag coef
foreach lag of numlist 1 4 6{
    regress close L`lag'.volume
    frame post results (`lag') (_b[L`lag'.`volume'])
}
frame results{
    twoway qfit coef lag, sort ytitle(Coefficient)
}
frame drop results

[Maxence Morlet]

As usual Andrew, your response is very helpful & instructive! Thanks! I will try to apply and let you know of the outcome.

[Maxence Morlet]

#2 worked perfectly, thanks again!