[R] SUR and less observation than parameters

Linus Holtermann holtermann at hwwi.org
Wed Oct 23 15:50:33 CEST 2013


   i  like  to  incorporate  a  SUR (Seemingly Unrelated Regression) in a
   multilevel-model. My panel-dataset includes variables for 400 regions from
   1990-2010.  In  case  of  SUR-models i found the following econometric
   requirement that time observations need to exceed unit observations, which
   is obviously not true in my dataset.
   Is it possible to avoid this requirement by using MCMC inference? I am given
   to understand that one advantage of MCMC estimation is due to the fact that
   the number of parameteres to be estimated can be larger than the observation
   number (because estimation does not rely on degree of freedoms). Can someone
   confirm my thoughts? In my case the discrepancy between time observations
   (20) and unit observations (400) is quite large. Is it possible to use a
   SUR_MCMC-model to handle this problem?

   Thanks in advance

   Mit freundlichen Grüßen

   Linus Holtermann

   Hamburgisches WeltWirtschaftsInstitut gemeinnützige GmbH (HWWI)
   Heimhuder Straße 71
   20148 Hamburg
   Tel +49-(0)40-340576-336
   Internet: [1]www.hwwi.org
   Email: [2]holtermann at hwwi.org

   AmtsgerichtHamburg HRB 94303
   Geschäftsführer: Prof. Dr. Thomas Straubhaar, Gunnar Geyer
   Umsatzsteuer-ID: DE 241849425


   1. http://www.hwwi.org/
   2. mailto:holtermann at hwwi.org

More information about the R-help mailing list