MIME-Version: 1.0 Content-Location: file:///C:/EC2AA333/Explaining_8.2,_8.3_revs.htm Content-Transfer-Encoding: quoted-printable Content-Type: text/html; charset="us-ascii" Explanation of the January 2004 revisions of Figure 8

Explanation of the January 2004 revisions of Figure 8.= 2 and Table 8.3

 

 

   &nbs= p;        These two revisions spring from improvements in the 1978-1995 data set and in the equation specification.  <= /span>

The improvements in the data related to a few years’ social transfer estimates in three countries = (Austria, N= orway, and Switzerland). 

The improvements in the reg= ression specification related to fixing the difficult conversion of predicted values from the first stage of the GLS regressions.  SHAZAM gives “observed”= ; and “predicted” values adjusted for heteroskedasticity and for seri= al correlation.  From these SHAZAM values, one must back out the raw predicted values of endogenous variables,= by reversing a procedure by Jam Kmenta (1986).  I mishandled the procedure in some= cases in the estimates of early 2003. 

The improvements have alrea= dy been incorporated into Volume 2 -- specifically, into Appendix E and Chapter 17.  (The estimates for Chapte= r 18 are unaffected.)  But there wa= s not time to correct Figure 8.2 and Table 8.3 in time for the first printing of Volume 1. 

I apologize for any inconvenience. 

 

--  Peter Lindert, 2 March 2004=