i t d rr,- r viSr-( waia.Ain Teniejp Mfrs


.,
 -I


Q oil
 Hr




 QI,, I


 QI ~-I


Q2.C? I IT
Q121




(r i


L1


B IA' B; B; Ai


A = :






 pll


 3



B;= :
 I '
 LJ


II
Iit,

o~il



Prill






Rin


0



0


-VOr
















 0I
 Ui


 The final matrix, E. consists of the distur-
bance terms re, with Elf.. = 0 for all -tliin-
tions i I .. n and all observations r 1,
... T. The covariance matrix of E is


 :j cr cr,
iri> V.iiKT-.1 r /-

 .C 0 r


where 1, is the Ti'rl-ordr idnciriL matrix."
 The earlier two equations -peitic.ltion is
clcarl% a special case of this Lcnerarl r'p.ii-
cation, The matrix represents the parameters
relating current prices of goods to each other.
The [, matrix is the set i.11 paraeters Iie all
prices lagged one time periMo: there are ad-
ditional ,B matrices up through Bl correspond-
ing 1 to the number of lagged prices included in
the equations. The As, again with as many pa-
rameter matrices as there are L.i, in the qu.ll-
tity variables, are di:igonil matrices since only
own lagged quantities are included in each

 The NpccLtihIlion assumes the variables
are stationary, In the absence of 'lt(iil,nrill.
the most likc-Ir scn.ilrii is variables inte-
grated ,t order I, Then ihl notion of the
price variables moving loticilici over time. ul
course, would be dealt with in the context oii
Lrinller.ilion Hwvt'eLr. our experience with
this approach for .irriicuilirl products has
not Iound nonstationarity ,ti the variables to
be an issue. Similarly, we have found only
lir.1 order lags of the variables to hr or in-
terest, In the :|hs-v .pec-itc.'IIin. that leaves
,ill,, H, among the B matrices, and 1 among
the A matrices.
 At first ighi. examination ,' like product
issues appears highly problematic with this
iiotdcl itcii the iii |i, po '-itl'ic combina-
tions. However, in the context it trade dis-
putes involving like product issues, piodut l,
can gfiierailly be separated into two groups
of commodities. The question then becomes
whether or not commodities in giIuIp I are
lihk products w itlh .,ImJinIIdilie, in gloup II7.
Assume for the moment lhal only lirst order


 hii s..pL.IIILJiFlIn assumeB no serial orrelation
that one would want to examine and adjust the spcci-
ticiiotn accordingly.
 Il ylls. one of the two groups will consist of
only one commodity: the one whose source is charged
with dumpming


-- -------------~ ~ -- ~-----