[R-SIG-Finance] EWMA covariance matrix

[Eric Zivot]

Does anyone have a suggestion for the best function to use to compute a EWMA covariance matrix? I see that the function cov.wt() (in stats) and cov.shrink (in corpcor) will compute a weighted covariance estimate. I assume I can just use exponentially declining weights as follows

> set.seed(123)
> testData = matrix(rnorm(100), 50, 2)
> lam = 0.9
> i = 0:49
> ewma.wt = lam^i
> ewma.wt = ewma.wt/sum(ewma.wt)
> cov.ewma = cov.wt(testData, wt=rev(ewma.wt))
> cov.ewma
$cov
            [,1]       [,2]
[1,]  0.8374426 -0.1366568
[2,] -0.1366568  1.0205296

$center
[1] 0.0456912 0.3136311

$n.obs
[1] 50

$wt
  [1] 0.0005756082 0.0006395647 0.0007106275 0.0007895861 0.0008773179
  [6] 0.0009747976 0.0010831085 0.0012034538 0.0013371709 0.0014857455
[11] 0.0016508283 0.0018342537 0.0020380597 0.0022645107 0.0025161230
[16] 0.0027956923 0.0031063247 0.0034514719 0.0038349688 0.0042610765
[21] 0.0047345294 0.0052605882 0.0058450980 0.0064945534 0.0072161704
[26] 0.0080179671 0.0089088523 0.0098987248 0.0109985831 0.0122206479
[31] 0.0135784977 0.0150872197 0.0167635774 0.0186261971 0.0206957746
[36] 0.0229953051 0.0255503390 0.0283892655 0.0315436284 0.0350484760
[41] 0.0389427511 0.0432697234 0.0480774705 0.0534194116 0.0593549018
[46] 0.0659498909 0.0732776566 0.0814196184 0.0904662427 0.1005180474


****************************************************************
*  Eric Zivot                  			               *
*  Professor and Gary Waterman Distinguished Scholar           *
*  Department of Economics                                     *
*  Adjunct Professor of Finance                                *
*  Adjunct Professor of Statistics
*  Box 353330                  email:  ezivot at u.washington.edu *
*  University of Washington    phone:  206-543-6715            *
*  Seattle, WA 98195-3330                                      *                                                           *
*  www:  http://faculty.washington.edu/ezivot                  *