*You must have coefficients and standard errors for separate regressions already stored. The graph code here corresponds to the scatterplot of coefficients and standard errors for bureaucratic quality shown on the website.
*Generate 95% confidence reference lines
gen posx = (_n/10)
replace posx=posx-0.1
gen negx=-(posx)
gen posrefline95=posx*(1/1.96)
gen negrefline95=negx*(1/1.96)

*Produce the 'V' graph. Remember to change mlabel(country) to the variable name which includes whatever characters you wish to label the points on your graph and 'sebur2' and 'betabur2' to the variable names containing the standard errors and coefficients from your stored regressions.

twoway (scatter sebur2 betabur2, msymbol(circle) mcolor(black) msize(small) mlabel(country) mlabcolor(black)) (line posrefline95 posx if posx<=10, lpattern(dash) lcolor(black) lwidth(medthick)) (line posrefline95 negx if negx>=-10, lpattern(dash) lcolor(black) lwidth(medthick)), title("Scatterplot of coefficients and SEs on bureaucratic quality", color(black) margin(medsmall)) subtitle("Post Cold War", color(black)) ytitle("Standard Errors", color(black) margin(small)) xtitle("Coefficients", color(black) margin(small)) legend(off)