MatchingLBO.do

From edegan.com
Revision as of 17:43, 25 October 2016 by Yimengtang (talk | contribs)
Jump to navigation Jump to search

This do file aims to match companies with the closest propensity score based on industry, year, and number of patents from STATApredictLBO.dta.

Attempt 1

use STATApredictLBO.dta, clear
rename nopatents case_nopatents
rename logitpw case_logitpw
rename gvkey case_gvkey

/*industry, year, and no. of patents*/
joinby industrygroup3 year using STATApredictLBO.dta
gen a = nopatents*0.95
gen b = nopatents*1.05
gen delta=abs(logitpw-case_logitpw)
drop if case_nopatents > b | case_nopatents < a

// GENERATE SOME RANDOM NUMBERS TO RANDOMLY BREAK ANY TIES
// ON CLOSENESS OF MATCH OF MARKET VALUE
set seed 0.5
gen double shuffle1 = runiform()
gen double shuffle2 = runiform()

by case_gvkey (delta shuffle1 shuffle2), sort: keep if _n == 1
drop delta shuffle1 shuffle2

Error op. sys. refuses to provide memory

Attempt 2

foreach x of varlish matchfilter{

 if x==1{
    foreach matchfilter[_n]==0{
       if year[_n]==year[_x] & industrygroup3[_n]==industrygroup3[_x]{
          if nopatents[_n]/nopatents[_x]<1.1 | nopatents[_n]/nopatents[_x]>1/1.1{
          gen delta=abs(logitpw[_n]-logitpw[_x]
          m=gvkey[_min(delta)]}
 }
 }
 }
Error invalid syntax