tag:blogger.com,1999:blog-6288862798546085706.post5366656065767513213..comments2023-01-31T03:23:26.561-05:00Comments on Econometrics By Simulation: Generate rank correlated variablesFrancishttp://www.blogger.com/profile/16658586705916884436noreply@blogger.comBlogger2125tag:blogger.com,1999:blog-6288862798546085706.post-4078830580844483322016-03-26T07:39:52.390-04:002016-03-26T07:39:52.390-04:00This is a very helpful post. If it is still possib...This is a very helpful post. If it is still possible that someone might answer, has anyone successfully used this to generate a beta distributed random variable from one of the correlated normal random variables? (I am clearly not industrious enough!) I have managed to obtain exponential random variables this way but I am struggling to get a beta random variable for a distribution that looks a little like an exponential would for a small mean value (<5).Puzzlednoreply@blogger.comtag:blogger.com,1999:blog-6288862798546085706.post-76049977434215253732015-10-02T15:19:28.338-04:002015-10-02T15:19:28.338-04:00I believe Mathworks has an analytical solution to ...I believe Mathworks has an analytical solution to generating a spearman rank. The general procedure is:<br />1) look at the correspondence graph/equation to computer the bivariate normal correlation needed for the spearman rank.<br />2) generate the bivariate normal (y1,y2)<br />3) if you need a distribution other than the bivariate normal (e.g. uniform), draw N readings of your distribution (z1) and perfectly rank-order correlate it with y1. Do this with y2 as well. As long as rank is preserved, the rank correlations should be the same.Anonymoushttps://www.blogger.com/profile/07792274390988577220noreply@blogger.com