Separate Appendix to:
NONPARAMETRIC COINTEGRATION ANALYSIS
by Herman J.Bierens
Following Phillips (1987), we use throughout this appendix the symbol "⇒" to indicate weak
convergence (cf. Billingsley 1968), convergence in distribution, or convergence in probability.
From the context it will be clear which mode of convergence applies.
Proof of Lemma 1: Denoting the partial sums associated with vt and wt by
[χn ]
Snv(χ) = 0 ifχ ∈ [0,n ^1); Snv(χ) = ∑vtif χ ∈ [n ^1,1L
t= 1
(A.1)
[xn ]
Snw(χ) = 0 ifχ ∈ [0,n l) Snw(χ) = ∑ft ifχ ∈ [n^1,ι].
t=1
respectively, it follows easily that
( A
Snv
vf
Sn
V √
D (1)J
W,
(A.2)
where W is a q-variate standard Wiener process. Next, denote the partial sums associated with
zt and ∆zt by
[χn ]
Sn(χ) = 0 if χ ∈ [0,n 1); Sn(χ) = ∑-χ if χ ∈ [n ^1,1],
t= 1
[χn]
(A.3)
s^δ(χ) = 0 if χ ∈ [0,n 1); Snz(χ) = ∑∆ztif χ ∈ [η ^1,1L
t=1
respectively. Then it follows from (3) and (A.2) that
37