Growth Rates

[Statanoob]

I have a huge panel data set consisting of thousands of firms for a time period of 8 years. I would like to transform some of my variables. For examples I have sales in USD and would like to calculate the SalesGrowthrate from that. I tried d.sales/l.sales and it seems that that respects the groups/firms so it could work. But when I look at the mean, sd, and Min max values they are to huge to be correct. And I already have a Growthrate variable in home currency which should at least be closely correlated even in respect to the exchange rates, but it is not.

I would like to do the same for another variable where I don’t have a control but the results don’t seem to be correct.

Is there another way to get the growth rate?

[ichbin]

The growth rate of X is not X_final / X_initial. It is either log(X_final / X_initial) / time or (X_final - X_initial) / X_initial / time, depending on whether you want continuous or one-time compounding.

[Statanoob]

That is why i used d.sales/l.sales

correct me if i am wrong but i though d.sales was first difference and l.sales the lagged value, so this would give me (X_final - X_initial) / X_initial where the time difference is one year. I dont like using the continuous transformation for one year differences.

I should be able to use (sales/l.sales)-1
But if look at the summary the min value should not be less then -1.
Im afraid that the groups get mixed up.