syntax help: to generate variables based on an equation

[Josh Wang]

Dear Statalisters,

Hope my post finds you well.

I need to generate two variables based on the equation: Ln(REVENUE_t)= b₁ + b₂t + ð where Ln(REVENUE_t) is natural logarithm of a firm's revenue in year t, t is the independent variable (year), and ð is residual.

First, run a linear regression model of the relationship between the year (t) and the natural logarithm of revenue (REVENUE) within the annual window period [t, t+4],

then I want to generate two variables.
1) X1: the antilog of the regression coefficient (b₂);
2) X2: the antilog of the standard deviation of b₂


Thank you in advance!

Best,
Josh


Here is the example of the sample:
* Example generated by -dataex-. For more info, type help dataex
clear
input id year REVENUE
1 2010 1000
1 2012 1200
2 2010 0
2 2011 900
2 2012 100
2 2013 500
2 2014 990
2 2015 1000
2 2016 1100
2 2017 1300
2 2018 1450
3 2010 1200
3 2011 1345
3 2012 0
3 2013 1340
3 2014 1250
3 2015 1400
3 2016 1568
3 2017 1200
3 2018 1900
4 2010 0
4 2011 0
4 2012 0
4 2013 400
4 2014 670
4 2015 760
4 2016 560
4 2017 800
4 2018 890
5 2010 110
end
[/CODE]
------------------

[Clyde Schechter]

I don't understand what you are asking for. Your equation is written for a single time series. But your data is panel data. I'm going to guess that you want to do these calculations separately for each id--which is really the only way it would make sense to have a variable to contain the results. Also, regression coefficients do not have standard deviations; they have standard errors. Of course, perhaps you mean to take the standard deviation of the different values of the coefficient across the different id's. I can't really tell what you want.

Anyway, my best guess as to what you are looking for is this:

gen ln_revenue = log(REVENUE)
rangestat (reg) ln_revenue year, by(id) interval(year . .)
gen X1 = exp(b_year)
gen X2 = exp(se_year)

-rangestat- is written by Robert Picard, Nick Cox, and Roberto Ferrer. It is available from SSC.

If this is not what you meant, please post back with a clearer explanation.

[Nick Cox]

#1 looks like an assignment to me.

If so, then please note our policy at #4 of https://www.statalist.org/forums/help#adviceextras

If not, then Clyde Schechter's reply in #2 above gives good advice as always.

[Josh Wang]

#1 looks like an assignment to me.

If so, then please note our policy at #4 of https://www.statalist.org/forums/help#adviceextras

If not, then Clyde Schechter's reply in #2 above gives good advice as always.

Hello Nick Cox ,
Thanks for your reply. It is not homework, I am doing a paper and now generating a variable used in some references.

Best,
Josh

[Josh Wang]

I don't understand what you are asking for. Your equation is written for a single time series. But your data is panel data. I'm going to guess that you want to do these calculations separately for each id--which is really the only way it would make sense to have a variable to contain the results. Also, regression coefficients do not have standard deviations; they have standard errors. Of course, perhaps you mean to take the standard deviation of the different values of the coefficient across the different id's. I can't really tell what you want.

Anyway, my best guess as to what you are looking for is this:

gen ln_revenue = log(REVENUE)
rangestat (reg) ln_revenue year, by(id) interval(year . .)
gen X1 = exp(b_year)
gen X2 = exp(se_year)

-rangestat- is written by Robert Picard, Nick Cox, and Roberto Ferrer. It is available from SSC.

If this is not what you meant, please post back with a clearer explanation.

Thank you Clyde for your codes. Sorry for making the confusion. It is a variable generated in a paper I cited. You can find here
https://www.emerald.com/insight/cont...0239/full/html

On page ten, the authors described the operationalization of the variables.

I have tried your syntax, it is working. But I'm just curious about the "within the annual window period [t, t+4]", this is indicated in the paper, the authors used a dataset from 2008 to 2017, but used this 5-year window period, how can we include this in the syntax?

Best,
Josh

[Clyde Schechter]

Oh, sorry, I forgot about the [t, t+4] window. Change the -rangestat- command to

rangestat (reg) ln_revenue year, by(id) interval(year 0 4)

[Josh Wang]

Oh, sorry, I forgot about the [t, t+4] window. Change the -rangestat- command to

rangestat (reg) ln_revenue year, by(id) interval(year 0 4)

Thank you for your help!

Best,
Josh