Natural log function

[Olena Onishchenko]

Dear Statalist members,

I am running simple code to calculate stock return:

generate var1return=log(var1 /L.var1)

Suppose var1=2080 and L.var1=1980.

var1return=0.04927105084

When I replicate the same prosedure in excel var1return=0.04927104901

Difference is 0.00000000183. My target variable is mean return.The difference between return estimated in Stata and Excel gets bigger for a big sample.

Are you aware of the reason for that difference?

[Nick Cox]

Use a double variable if you want more precision. Stata can echo that result; the problem is that your own generate command produces a float.

. di %23.18f   ln(2080/1980)
   0.049271049006782835

[Olena Onishchenko]

Thank you Nick. It works perfectly now.