* compute average over time by firm
* use collapse commands
collapse (median) sale (median) xsga , by(gvkey)

gen ln_s = ln(sale)
gen ln_xs = ln(xsga)

xtile k=ln_s, nq(10)

mat mat_results = [.,.]

forvalues j=1/10{
reg ln_xs ln_s if k == `j'
mat mat_results = mat_results \[ ( _b[ln_s], _se[ln_s])]

}
matlist mat_results

svmat mat_results

keep if mat_results1 ~=.

gen index = (_n)*10


* Elasticity of SGAX with respect to sales
graph twoway (scatter mat_results1 index),xtitle('Percentile %') ytitle('Elasticity')xlabel (0 (10) 100)


怎么在图中加入95%confidence interval? 如果用gen UB=_b[ln_s] + 1.96*_se[ln_s]
                                                                         gen LW =_b[ln_s] - 1.96*_se[ln_s]这应该加到哪呢？
谢谢