Email discussion lists for the UK Education and Research communities

## allstat@JISCMAIL.AC.UK

#### View:

 Message: [ First | Previous | Next | Last ] By Topic: [ First | Previous | Next | Last ] By Author: [ First | Previous | Next | Last ] Font: Monospaced Font

#### Options

Subject:

From:

Date:

Mon, 12 Apr 1999 12:56:15 +0100

Content-Type:

text/plain

Parts/Attachments:

 text/plain (33 lines)
 Any ideas on this - a friend of mine wanted to know this. A transition probability matrix gives us the probabilities with which one income state has the possibility of transferring to another state. In the light of understanding transitional intra-distributional dynamics of , say, income levels across regions, such a matrix ( or a stochastic kernel) gives us the probabilities with which regional incomes transfer from one state to another- hence, if the diagonals have high probability values, that is an indication of little intra-distributional mobility over time. Thus regional distribution of incomes have not changed over time. How would it be possible for us to determine the following : what determines why state A after a certain period moves into state B? How could we find out what determines the probabilities One method which is used is that of conditioning- seeing whether E(y) and E(y/x) are different (i.e. the distributions)- if so, we can say that factor x has been instrumental in transforming the distribution of y (over time period ,say s) Thus one can find that different factors, such as investment etc. have been instrumental in determining the intra-distributional dynamics. This is the method traditionally used in regression analysis as well. Is there any other way of determining what determines why state A moves into state B over time period t to t+s? Regards, Sankarshan Basu. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%