Tilburg University
How sensitive are average derivatives?
Härdle, W.K.; Tsybakov, A.B.
Publication date:
1994
Document Version
Publisher's PDF, also known as Version of record
Link to publication in Tilburg University Research Portal
Citation for published version (APA):
Härdle, W. K., & Tsybakov, A. B. (1994). How sensitive are average derivatives? (Reprint Series). CentER for
Economic Research.
General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners
and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.
• Users may download and print one copy of any publication from the public portal for the purpose of private study or research.
• You may not further distribute the material or use it for any profit-making activity or commercial gain
• You may freely distribute the URL identifying the publication in the public portal
Take down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately
and investigate your claim.
CBM
R
for
~a,s, ~`~~"n III~nIIIIIIIIIIIIIIIIIIVIIIIIQIIINIhllllllll
How Sensitive are Average
Derivatives?
by
Wolfgang H~rdle and
A. B. Tsybakov
Reprinted from Journal of Econometrics,
Vol. 58, 1993, North-Holland
,J~,~~~
Reprint Series
CENTER FOR ECONOMIC RESEARCH
Board
Ham' Barkema Helmut Bester
Eric van Damme, Chairman Frank van der Duyn Schouten Jeffrey James
Management
Jeffrey James (Director of Graduate Studies) Arie Kapteyn (Scientific Director)
Marie-Louise Kemperman (Managing Director)
Scientific Couocil Anton Barten Eduard Bomhoff Willem Buiter Jacques Drèze Jack Kleijnen
Theo van de Klundert
Jean-Jacques Laffont Merton Miller Piet Moerland Philippe Naert Pieter Ruys Residential Fellows Hans Bloemen Lans Bovenberg Hans Carlsson
Jay Pil Choi
Jan Magnus Andrew Mountford Bezalel Peleg Mark Steel Frank Verboven Oscar Volij Karl-Erik Wtimeryd
Université Catholique de Louvain Erasmus University Rotterdam Yale University
Université Catholique de Louvain Tilburg University
Tilburg University
Université des Sciences Sociales de Toulouse University of Chicago
Tilburg University Nijenrode University Tilburg University CentER
CentER, Erasmus University Rotterdam Gothenburg University and Lund University Columbia University
CentER, LSE CentER
Hebrew University of Jerusalem
CentER~Department of Econometrics, Tilburg University CentER
Hebrew University of Jerusalem S[ockholm School of Economics
Address : P.O. Box 90153, 5000 LE Tilburg, The Netherlands
How Sensitive are Average
Derivatives?
by
Wolfgang Hárdle and
A. B. Tsybakov
í;-
ii
J~l~~'~ a~Li-'.~..tc.Í ' ~ ~;i~P..1 . ~
~~
TiLái~h~à
Reprinted from Journal of Econometrics,
Vol. 58, 1993, North-Holland
Journal of Economelrics 58 (19931 31--48. North-Holland
How sensitive are average derivatives?~`
Wo(fgang Hárdle
Tilburg UrrioerxitJ~, 5000 LE Tilburg, The Netherlnnd.r
Unirersité Cnrholique de Louuain. B-l348 Louuain-lo-Neuue, Belgiunr
A.B. Tsybakov
Tilburg Unirersify~, 5000 LE Tilburg, The Nclherlonds
Average derivatives are the mean slopes of regression functions. In practice Ihey are estimated via a nonparametric smoothing technique. Every smoothing method needs a calibration parameter that determincs the finite sample performance. In this paper we use the kernel estimation method and develop a formula for lhe bandwidth that describes the sensitivity of the average derivative estimalor. One can determine an oplimal smoothing parameter from this formula which tries out to undersmooth the density ofthe regression variable.
1. Average derivatives in discrete choice analysis
The average derivative is the mean of the slope of a regression function. In
a regression setting Y- m(X ) -F- e with regression curve m: R" -~ R, the average
derivative is the mean gradient Ex(m'(X)), or, more generally, the weighted
mean gradient
á - Ex(m'(X)w(X)),
(i.l)
where ní (x) is the gradient
am
am
m'(x) -
aX ,...
ax )
e R",
t e
x,, ..., xa are components of the vector x, w(x) is some weight function, and
Er is the expectation with respect to the (marginal) X-distribution.
Correspondence ro: Wolfgang Hfirdle, fnstitut fiir Statistik und dkonometrie, FB Wirtschaflswissen-schaften, Humboldt-Universitiit zu Berlin, D-1020 Berlin, Germany.
'Work of the second author was financially supported by the Department of Econometrics, Tilburg University. The Netherlands.
32 II'. HnrJh rr~ul ,~.B. T~rhuknr. Hnu~ sensi~ire nrr nceraRe rlerirniiuaa?
The average derivative ó is interesting in the context of discrete choice
analysis, where in the case of binary choice we want to infer on the
function
P( Y- I I X-.Y) - Ill(X),
from observations {(X;, Y;)};: ~, X; e R', Y; e {0, 1}. A pure nonparametric
approach to estimation of m(x) is possible [see, for example, the recent
mono-graphs by Muller(1988), Eubank ( 1988), Wahba ( 1990), and Híirdle(1990)]. It is
well-known though that this approach is not costless: the precision of the
estimator is exponentially decreasing as the dimension d increases. In order to
avoid this difTiculty one could of course fall back into pure parametric models
for m(.r).
-One such model would be
pl(X) - G(Yr~), (1.2)
where G, the link function, is of known form, e.g., G-~ would postu)ate
a Probit model.
A model comprising the advantages and simplicity of (1.2) and the flexibility
of a nonparametric smoothing approach is a single-index model,
!II(.Y) - lIl-YTl'), (1.3)
with an unknown link function y and index xT(i.
It is well-known that (i in (1.3) can only be identified up to scale [see
Hárdle and Stoker (1989)]: the (weighted) average derivative (ADE) for this
modef is
b- Ex
L
d(XT~) w(X)
J
l~ -)'r'í~,
(1.4)
so we see that we can estimate ~3 (up to scale) if we know how to estimate S and if
~~B is difTerent from zero. A simple examp)e for (1.4) is a linear link function g(' );
then the coefiïcients ~ are multiplied by the slope of g(-) times EX(w(X)). For
general, nonlinear g( .), as in binary choice models, the ~3 coe(Ticients are
multiplied by the average slope
d
!V. Nnrdle mid A.B. T.~rbaknr, Ho~~ s~.nsrrine nre nrernge derivarrnes? 33
We use kernel estimators for the average derivative b since they are
straiglit-forward to implement and easy to understand on an intuitive Ievel. Other
possibilities include splines and
orthogonal series, but to our
know-ledge these techniques have not been employed to estimate average
deriva-tives. The main point in this paper is about the selection of the bandwidth,
the kernel smoolhing parameler, for lhe d-dimensional case. The
one-dímensional case with a focus on estimation of income efïects is treated in
H~rdle, Hart, Marron, and Tsybakov ( 1991). From an asymptotical viewpoint
the choice of bandwidth does not afiect the behavior of ADE estimators.
It influences only the higher-order terms of asymptotic expansions for mean
squared error, not the main term which is of order O(I~n), where n is the
number of observations. In practice though, the choice of the smoothing
parameter is an important issue as has been pointed out by Hsieh and Manski
(1987, p. 55I).
In this paper we consider tlle special choice of weight function: w(x) - j(x), where j(.Y) is the marginal density of X[cL Powell, Stock, and Stoker ( 1989)].
This is motivated by several reasons. First, under such choice of w we avoid tlie
random denominator appearing if rv(x) - I [in fact, for w(x) - I the ADE estimators contain the density estimator in denominator, see Hiirdle and 5toker (1989) for details]. Because of lhe random denominator the necessary asymp-totic expansions hold under somewhat restrictive assumptions on the underly-ing density j [H~rdle and Stoker ( 1989), Hiirdle, Hart, Marron, and Tsybakov (1991)]. Next, for the multi-dimensional case the O(I~n) rate of the mean-squared error is nol attained unless the oscillating higher-order kernels are implemented. This causes a difliculty in trealing the case of w(x) - I: the ADE estimator is not well-defined and it requires some truncation [Híirdle and
Stoker ( 1989)]. The choice of truncation threshold appears to be crucial in this context. This creates an additional problem which could be easily eliminated if
ll'(.Y) - j(X).
In section 2 we quantify the sensitivity of ADE via a second-order expansion
of inean squared error of a kernel estimator for b. Section 3 is devoted to the
proof ot our main theorem. [n the appendix we prove some lemmas.
2. The sensitivity of ADE
Assume tliat independent pairs (X;, Y;), i- 1, ..., n, of random variables,
X; e R", Y; e R', are observed and that they have the same distribution as
(X,Y),XeR",YeR'.
Let the regression function m(.Y) - E( Y~X - x) exist and let X have the
density j(x) wilh respect to Lebesgue measure in R'. Suppose, moreover, that the
regression function rn and the density jare continuously difierentiable and that
34 1{'. Hiirdle ain! A.B. T.r~hakor, Nmr .rr~tririve nre nverage deriralioe.r?
Using partial integration (over the support of X) we get
b - J m'(x)J'(r)dx
- - 2 Jm(x)f'(x)f(x)dx
- -2E(YI~(X)).
wliereJ
'(.r) -(a~~ax,
, ..., af~a.rd) and tlie expectation is now taken over the
joint distribution of (X, Y).
If we knew the marginal density Jwe could estimate S by means of the sum
-(2~ri)~~-r Y;J'(X;) which is obtained if one substitutes the expectation in
(2.l) by the empirical average.
In our approach we do not know the density function. We shall estimate it
from the data via the kernel method. The marginal densityJ(-) is estimated by
n
Jh(-C) - n-~ ~ .~h(x - Xi),
i-l
where .~Yh(u) - h-",~Y'(ul~h, ..-, nd~h) for a multivariate kernel function, e
.X'(u, , . . . , ne) - ]-] K(u~).
u - (u, , . . . , ue) E Re,
(2.3)
i-r
based on a one-dimensional kernel K. The scaling of .7Y~ is through h~ 0, the
bandwidth, or smoothing parameter.
The gradient J'(x) is estimated by
1
h(x) -' G., ~h~Y - Xi),n;-i
where
~n(lt) - h-e-' K' I uj~ n K~u'~,
`h
k:j
h
and K' denotes the derivative of one-dimensional kernel K.
Using (2.4) we can construct an estimate of the average derivative
b~ - - - ~ Y~fí,(X~).
(2.5)
I{'. Nnrdle~ iuul A.R. T~rhakat', Huu srnsr,ioc arr ne~raRe derioariues? 3S
We study the asymptotic mean squared error of b„ under the following
assump-tions:
(A I) The kernel K is bounded, continuously difïerentiable, symmetric with
support [ - I, I]; K'(0) - 0.
(A2) JK(u)du - I, and there exists a positive integer k~ 2 such that Ju~ K(u)
xdu-0, j- I,...,k- I, f ukK(rQdu-dK~O.
(A3) The marginal density f(.~) of X is compactly supported and has eontinuous
partial derivatives up to the order k f I on Rd.
(A4) The regression function rn(x) has continuous partial derivatives up to the
order k f I on Rd.
(AS) The conditional variance aZ(x) - var( Y~X - x) is bounded on the
sup-port of f.
(Atí) h-h„-.O,andnZh~tZ-. oo asn--~ oo.
Later ~.~ denotes Euclidean norm when applied to vectors.
Theorem.
WIIPPP
Clnd
Under the as.u,niptiorrs (AI)-(A6),
E(ló~ - álZ) - Q,~~-~
f Q2~r-Zl~n d-Zt Qah~k
k ~-t'~l't~~r'Z~fO r'Zh„t2 }ll~kQ, - 4[E(IÍ(X ) n~'(X)~Z) - IE(Ï(X)nt'(X))IZ
f E(QZ(X)II~(X)I~)],
Q2 - 4CK J OZ(X)f 2(C)dx, Q3 - 4I J SK (-Y)I(x) Rl(X) dX I z,C„ - J ~.~Y"(u)~2 du - d f(K'(u))Z du ( J K 2(u) drQ"- `,
( - 1 )k d
ak } ~ .t(c)~aX, ax;
SK(-~) - dK ~ :
k!
j-~ ak.r f(C)~aY,ar;
36 {P. H~irdle and A.B. T.~7'h~lkur, Hnn'si~~esinr~t' arr' are~nke rlerirurir~~s?
From the Tlteorent we see tl)at the bandwidtlt ll„ minimizing E(~b„ - b~z) is
given by
h' - It rt-zrrzktd.zl
„
o
where
Ito - (Qz(ri
f 2)1'nzk,e.z)
`
2kQ~
J
For h„ - h~ , we have
E(Ib„ - blz) - Q~n-' -i. Crl-4kl(2ktdf2) } O(q-4k112ktda2l) f ~ ~ ,2 ~, I7 ~ 00 , n
where
2k )dt2)I(2kfdt2) d } 2 2kllzktdt2)C-
df2)
}( 2k )
)
xQ2kl12kfdf21Q3 f21f12kidtzlOprirnization oj k.
This in fact is reasonable if one believes that f and rn are
infinitely many times continuously difTerenliable. It follows from (2.6) that the
best rate for mean squared error equals n-' and it is attained if k ~(d f 2)~2.
For example, in one-dimensional case ( d - 1) it sufiices to take k- 2. Then tlte
second term in (2.6) equals Cn-8r', and !t~ is proportional to n-zj7 [cf Híirdle,
Hart, Marron, and Tsybakov ( 1991)].
Assumptions ( AI) and ( A2) entail that the order k of the kernel should be
necessarily even. Thus, the condition for choosing k that guarantees the best rate
of convergence becomes:
k is Ure mininm( euen number such that k ~(d f 2)~2.
Optirrtization oj K.
The factor C depends on the kernel K. Optimizing this
factor in K leads to the minimization problem (in view of the definition of
Qz attd Q;):
min ( J uk K ( u) du)d ~ z( J( K'(u))z du)zk ( J K z(u) du )zrd- t)
N~. Hwdlc nrid A.B. Tsrhakor, Ho~r sensitrce ore auerage derieatines? 37
where ~Ih is the class of kernels satistying ( A l) and (A2). For d- I, this problem
was solved by Mammitzsch ( 1990) who showed that the optimal K is the quartic
kernel
K(u) - i~(I - uz)z 1(~u~ 5 1),
where I(. ) denotes ihe indicator function. If d ~ 2, then k ~ 4, and the optimal
K is, clearly, an oscillating kernel taking positive and negalive values.
3. Proof of the Theorem
Denote
ná~-ó"--1 S~ Y-
X.,
ó'-b-- mx
x {~ xdx.
2 ll.ri- ifh( ~) 2 J ()fI( )J( )Clearly,
E(~b" - b~Z) - 4E(~á~ - b'~2).
(3.1)
Write the estimator S` as
b~ - ~ ~ pn(Xt) f E,)Íh(X;),
t,;-,
where s; - Y; - m(X;). Since E(e,~X;) - 0, we have
E(ló~ - b`IZ)
~ Í
- E E;Íh(Xt)
,t;~,
1 nn;-t
- ~ m(Xr)ií~(Xt) - b~
1 n ~'~ E;Ïi~(Xt)IZ~ -f- E~I ~'~t CSr - E(l,r))1 n 2
f
n ; ~,
E(S,) - b'
where S; - m(X;)Jh(Xf).
38 li'. Hiv~lle anJ A.B. Tsrbukor, Hom sensiliue arc ntr~nge derifwlines?
It follows from (A1) that .~f~~,(0) - 0, and thus
1
Ih(Xf) - - ~, ~I,(Xf -
X;)-It j-, ixj
Thus,
E(Si) - E(Sr) - E(mlXr)Ií,(Xf )) -
n - 1
y,
(3.4)
n
where q - E(m(X,).JF~~,(X, - XZ)). Here we used ( 2.4) and the fact that
.7Y~,(0) - 0 which follows from (A1). Now, (3.2) and (3.4) entail
E(Ib~ - b'12) - Vr f vZ f v,,
where
I
~"
n,-' Effh(Xi)}
~ ~ (bf - E(~r))IZ~.
n r-,
v~ - IE(Cf) - b'~2.
Let us evaluate the terms Vl, VZ, and V,.
Clearly,
J~Z(Xi),
1- f.
E(EiEI~XI, .., Xn)-1~, i~ l.
From (3.3) and (3.6) we get
t6'. Hórdle mul A.B. Tst'hnkav, Hou~ srrtsiliue ore average rlerinarives? - ~~J ~~2(xt) E ~ 2
~ .~„(xt -x;) )j(x,)dx,
J-z
- ~~a
J
vz(x) E
~ ~ (,~f~í~(x - XJ),.N'~"n(x - X,))~Í(x)dx.
~:-z
Here and later (~,.) denotes the scalar product. We have
E ~ ~ (.7Y~,(x - XJ), ,~í,(x - X,))
J
`,, z
39
n
-~ E(I~h(x - x;)IZ) f~ IE(~~,(x - xJ))IZ
(3.s)
J-z J..-z
Js:
- (17 - 1) E(~~h(x - il 1 )~Z)
f (~i - 1)(~~ - 2)IE(~h(x - Xt))IZ.
[t follows from Lemmas 1 and 2 and (3.8) that
E ~ ~ (.~Y'í~(z - X ),~i,(.C - x,)~ `, . z
- (n -1) (~Xj(x)l~-'-Z -~ n3vl, x))
f(n - 1)(~i - 2)Ij'(x) -t- hk Sx(x) f Qt(h, x)~z.
Hence
Vt -
1
1
J
QZ(x)Ij~(x)IZ
I(X)dx i- n2Jtd~2 C~J
v2(x)j~(x)dx
1
~h4
1 i
~o
f0
- t zl,
n~ oo,
(3.9)
rtzh"'z
n
n
40 N'. Hn"rdle nnd A.B. Tsrh~tkno, Hmr .aensifirr are auerage dcri~atioes?
Next,
n
~z - 2~ E((~,,~;)) - IE(~t)Iz - ~zt -~ ~zz - IE(~,)Iz,
n ;.~-,
(3.to)
where
~
1il - Z~ E(Istlz),
n ;31
~zz - Z~ E((St, ~;)).
n ;.Ja,
t:~
We have
E(~Sriz) -'z E Intz(Xf) ~(~á(Xt - Xf),.~n(XI - Xs)) I.
n
I.sxt
Hence
1
"
l
Vzt - ~~3 E mz(XI) ~ (~í,(Xt - Xf), ~í,(XI - Xz))J.
f.:-z
Using the same argument as in (3.7)-(3.9), we find
V21 - i J
l112~~)~f (-~)~ZI(.Y)dX f n2I1 ~Z CK JnIZ(X)f Z(X)dX
1
kf o
nzli tz
J
f O 'n f nz '
n-~ oo.
(3.11)
Consider the term l 2z now. Applying (3.3) we obtain
11'. Ifiudl~ nnrl A.B. Tcrbaknr. Hm~ srn.airiee ure nrernge Arriralit~e.c?
where
dls - E[m(X,)rn(Xz)(-~ í~lXl - XI),.1f í,(Xz - X,))).
Let us treat d,,, separately in the following four cases:
(i)
s - I,
(ii) s~l,l~2,s~1,
(iii) s~! and either 1- 2, s~ I or 1~ 2, s- 1,
(iv) s~l,l-2,s- 1.
In case (i),
41 d,s - du - E[n,(XI)nl(Xz) (.~á(X, - X~), ~í~(X2 - X3))~(3.13)
def - E.r.~~lEx~(r,l(XI).~h(XI - X 3))I2~ - BI.In case (ii),
d,.. - E[rn(X, )n,(Xz)(~~i~(XI - XI), ~-í~(XI - Xs))~
- IE(In(X, ).~~n(X, - X,))12 -1912.
In case (iii),
dr. - E[m(X, )n~IXz)(.~h(XI - Xz), ~ n(Xz - X.)))
dcr
- E~(Ip(XI)lll(X2)(~h(XI - X2))..~h(X2 - X D))~ - B2.
lIl CasC (IV), .
dlt - E[,n(X,)nt(Xz)(~í~(XI - Xz), ~h(Xz - XI))7
-
-E ! n, X) n, X
~l ( I ( 2)~.JY' X
h( 1-X
2)~2) - 83-d~r
In (3.16) we used the fact that .X'~, is antisymmetric:
(3.14)
(3.15)
(3.16)
42 li'. Nàrdln ond A.B. T.crhakou, Hmr sensilii,e nre neernge rlericalinn.c?
It follows from (3.12)-(3.16) that
Vz2-Jl-1
n~ C(
n-28 f n2-Sn.f-6
) J(
)19I f 2(n - 2)BZ t B,].
2
(3. I 7)
Now we use Lemma 4 in the appendix, which implies, together with (3.17),
lhat
V22
-1LJ~j(Y)IJJI~(.Y)I2dX -~IA2(x)~f,(x)~2f(x)dX J
f 1912 -E ~1 - 6~ - n2lle.z fm2(x)I2(x)dx
f OC"k') C')
n f JJ2 f o n2lJ4~2 ,From (3.4), (3.10), ( 3.11), and ( 3.I8), we find
V2 VSJ i V22
-n-1
9
J z - 1'zJ f Vz2 - ~9~2 I 1-? f~2~ ` n n- n L Jf
~(x)~rJJ'(x)I2dx - 41912J
f 0 IJ } 1J2 ~ O n2l14~2 ' ClJk 1 I ~ , ~By substitution on (A.2) into (3.19), we obtain
12 - ' L
JÍ'(x)~Jn'(x)12dx - 41ó~~2J
n -. ao.
(3.I8)
(3.19)
k
n'. Hárdle nnrl A.B. Ts}~hnkor, Noir sensirice are auerngc derivalice.t? 43
tt remains to find the asymptotic expression for V3.
By (A.2) we have
V3 - n - 1 rl~~ llk
J
SAl-r)iGr) nl(z)dx f ~Iz(ll)
n
- 112k k -F 0 l~ f IZ t O(IIZk), II nq-b'
~ SK (-C) f(X) I11 ~Y) dX I2 z z I t ---~ ~ .From (3.5), (3.9), ( 3.20), and ( 3.21) one gets
E(Ib~ - h~`Iz)
- n L J f lx
(.Y)~rn'(X)Iz dx - 41b'~z f
J
az(X)II
~(X)~zÍ(X)dX
J
1
~- nzlíd.z Cx
az(X)Iz(X)dx f
hzk
J
.SK (X) f( X) nl(X) dX kfO~rlífnz~-Fo
nzli,z~'
rt~oo.
This, in view of (3.1), proves the Theorem.
2
(3.21)
Appendix
Lennnn l. Lel assumpirons (AI)-(A4) be sntisfied. Then
44 IV. Hirrdlc and .4.8. Tsrhnkor, Hu~r sensilive nrp ar~erngc derivalires?
~ahere sup,l~,(h, x)~ - o(hk) ns h~ 0, and
9- r5' - hk
J
SA(~)I(C)!ll(x)dx ~~2(Ir), (A.2)~c
here
I~iz(h)I - o(h4) as Ir ~0.
Proof.
By partial integration,
E(~h(Y-X,))-1idtl
J
.)f"~xlr z~f(z)dz
1
- - h f~'(u)I(x - Uh)du
(A.3)
1
- Ir (f(X - UJ7))',7Y(U)dlr-J
.~' (Y - lfh)JEr(ll)dU,where we used the fact thal .~Y and f are compactly supported. Assumption (A3)
entails that the Taylor expansion is valid, and thus, uniformly in u e supp.~Y~,
a~o~ t ~.t(x)~ar, ax'
f'(x - nrr) - ~ ~r (- I)~'~rr'Ir~'~
.
a:~a~ src a-
a
~a~ t rf(x)~axdax
aS Qr (h, x),
(A.4)
where supxl~i,(h, x)~ - o(Irk), h~ 0, and a -(a, ..., ad) is multi-index,
a;10, ~x~-a, f...~ad al-a~l...ad1 ,ra-u~~...uë for
u-(u, , . . , nd) e Rd, and a~'~~ax' - a~'I~aY ~~ . . . .. axá" for x - (x,, . . , .rd) e Rd.
It follows from (A2) that
J
rr'JY(tr)du - 0,
0 S Ix~ 5 k- I,
(A.5)
0,
~x~ - k, card { j: a; ~ 0} ~ I,
tr'JF'(tr)dn
IV. Hitrrllt~ und .i.R. T.ttbnAor, Notr sertsirit~r arr aonrnge deiirarloes? 45
From (A.3)-(A.5) we get
E(.~Y'h(.Y - X, ))
a~s~ } ~.f(.x)~aX, arQ
-- j 1-v) - ki Il"(- I)`
~
.
a:~al-k alal } 1j(.K),aXdaXa
x
J
u'.~Y(1Qdu-~l,(Il,x)
(A.6)
-- j'(x) - hk SK(x) - 1' 1(Il. X).
which proves (A.1). To get (A.2) note that by (A.I)
CJ - E(Ill(~il).it~hlX1 - X2))
- E(nl(x~t)Es:(~i~(Xt -Xz))
- EOn( Y, )(-j'(X 1) - hkS,;(X1) - ÍItUh X 1)))
- (5 ~` - hk
J
Sk(X)I(-Y) )n(X) dX - ~ITI(X) j(X) ~ÍI (I7, x) dX.Now, sup,~~i, ( h, x)~ - o(h~), and J m(x)j(x)d.x ~ oo since nl and jare bounded
on the compact support of f. Tliis proves (A.2).
Lemnra 1.
Let assurnplinns ( Al )-(A4) be satisjed. Then
E(I~ti(.x - Xr )I~) - CA j(x)h-"-Z f llallh-x),
ivJrere sup,.~~i(Il,.x)~ - o(Il"d"Z), Il-~0.
Proof. E( ~.~F'~h(.x - X, )I Z) - Itze.z f
I~ `x h Z~ z
j(Z) dZ
1
46 t1'. Hrvdle m~d A.B. T.rrhakov. Nmi~ sensirine are average derinariues?
It follows from (A3) that f is Lipschitz continuous on its support with some
Lipschitz constant Lf. Thus,
Ilrétz jl~'(u)Iz.Í(x - uh) du - li tz.f(x) Cxl S Irét, fI.~f"(u)Iz lul du.
This proves the lemma.
Len,mn 3. Let assumptions (A!)-(A4) he satisfied. Tlren
E(rn(X 1)~h(x - X1)) --(!n(x)f(x)) - IIkS1K(x) - l~d(Ir, X), lVll ere
ak}1 (f(x)nl(x)),aX~aXk
S,x(x) - dx (-
1)k ~
;
kr
;` ~
ak},
U(x)rn(x))~ax,ax;
and sup,l~3a(h, x) - o(Irk), h-. 0.
Proof.
The proof is similar to flie proof of (A-I ). In fact, instead of (A.3) we now
have
E(rn(X~
I.XI~,(x - X~)) -
J
U(-x - uh)m(x - uh))
'.JEr(u)du
-- f[f'(x - ul,) m(x - ulr)
~-fcx - ulr)rr,'
(x - uh)].~Y'(u)du.
Len,ma 4.
Let assumptions (AI)-(A4) be satisfied. 77ren, as h-~ 0,
B, -
J
~nl'(z)Iz!(x)dx f
J
mz(z)I.Í~(x)IZÍ(x)dx
-F 2
J
n,(x)(rn'(x),f'(x))J'z(x)dx f O(h4),
(A.7)
Bz - - ~nrz(z)II'(x)IZI(x) dx
-
J
m(x)(rrí (x),j'(x))f z(z)dx t O(h"),
(A.8)
I
r
fV. lldrdle mid A.B. T.c}~finkov. Nom sensirrne nre ai~erage deriualives? 47
Proof.
By Lemma 3,
Br -
J
I(x)I E(rn(X r)~í,(Xr - x))IZ dx
- fI(x)I(nr(x)I(x))' ~- h"Srx(x) -t- ~3s(x, h)IZ dx
- JI(r)I(nr(x)I(x))'Iz dx f O(hk),
which gives (A.7).
Using Lemmas 1 and 3 and the antisymme[ric property of .7Yh, we get
Bz
-J
m(x)m(Y)(~í,(x - Y), ~n(Y - z))I(x)I(Y)I(z) dx dy dz
-
J
f(1')nr(Y) 1J
~í,(x - Y)I (-z) m(x) dx, fI(z)~~(Y - z)dz~dy
-
J
I(Y)rn(Y)((nr(Y)I(Y))' i- h4Srrc(l') i- Í~a(Ir,Y).-.I'(Y) - hkS~(Y) - ~r(11,1'))dl'
- -
J
I(Y)rn(Y)((m(Y)J(Y))'.Í'(Y))dy.
This proves (A.8). To prove ( A.9) note that with ~v -(x - y)~h we have
B~ - ~rrt (x)nt(1')I(x)I(Y)I~n(x - Y)IZ dx dy
1
-- hi.z
nr(Y)I(Y)nr(y f wh)Í(Y ~- wlt)I~í,(w)Iz dydw.
It follows from assumptions ( A3) and ( A4) that j and m are Lipschitz continuous
on the support of f. Hcnce
B~ -- ha:: [ f rnz(Y)Iz(Y)dY J ~~í,(w)Iz dw(1 f O(h))J,
48 {I'. Hnrdle tmd A.B. T.crhnkot~, Hmr sertrilir~ are ooer~Re Jeritxniues?
References
Eubank, R., 1988, Spline smoothing and nonparametric regression (Marcel Dekker, New Vork, NY).
Hárdle, W., 1990, Applicd nonparametric regression, Econometric Sociely monograph series 19 (Cambridge University Press, Cambridge).
Hàrdle, W. and T. Stoker, 1989, Investigating smooth multiple regression by the method of average derivatives, Journal of the American Stalistical Association 84, 9R6-99S.
Hàrdle, W., l. Hart, J.S. Marron, and A.B. Tsybakov, 1991, Bandwith choice for average derivative estimation, lournal of the American Statistical Association 87, 218-226.
Hsieh, D.A. and C.F. Manski, 1987, Monte Carlo evidence on adaptive maximum likelihood estimation, Annals of Statistics I5, 541-SSI.
Mammilzsch, V., 1990, Asymptotically optimal kcrnels for average derivative estimation, in: Prottedings of the IOth Prague conference on statislical decision funclions, information theory and random processes, Prague, Aug. 1990.
Muller, H.G., 1988, Nonparametric regression analysis of longitudinal data, Springer lecture notes in statistics 46 (Springer-Verlag, New York, NY).
Powell, J.L., 1.H. Stock, and T.M. Stoker, Semiparametric estimation of index coefficients, Econo-mctrica S7, 1403-1431.
Reprint Series, CentER, Tilburg University, The Netherlands:
No. t G. Marini and F. van der Ploeg, Monetary and fiscal policy in an optimising model with capital accumulation and finite lives, The Econartic Jorvnal, vol. 98, no. 392,
1988, Pp. 772 - 786.
No. 2 F. van der Ploeg, International policy coordination in interdependent monetary
economies, Journa[ of Luernational Econornics, vol. 25, 1988, pp. I- 23.
No. 3 A.P. Barten, The history of Dutch macroeconomic modelling (1936-1986), in W. Driehuis, M.M.G. Fase and H. den Hartog (eds.), Cl:allenges for Macroeconomic Modellirtg, Contributions to Economic Analysis 178, Amsterdam: North-Holland, 1988, PP. 39 - 88.
No. 4 F. van der Ploeg, Disposable income, unemployment, inflation and state spending in a dynamic political-economic model, Public Choice, vol. 60, 1989, pp. 211 - 239. No. 5 Th. ten Raa and F. van der Ploeg, A statistical approach to the problem of negatives
in input-output analysis, Econontic Mode!ling, vol. 6, no. l, 1989, pp. 2- 19. No. 6 E. van Damme, Renegotiation-proof equilibria in repeated prisoners' dilemma,
Jounta! oj Economic 7lteory, vol. 47, no. 1, 1989, pp. 206 - 217.
No. 7 C. Mulder and F. van der Ploeg, Trade unions, investment and employment in a
small open economy: a Dutch perspective, in J. Muysken and C. de Neubourg (eds.),
Unernploymen[ in Europe, London: The Macmillan Press Ltd, 1989, pp. 200 - 229.
No. 8 Th. van de Klundert and F. van der Ploeg, Wage rigidity and capital mobílity in an optimizing model of a small open econotny, De Ecortwnist, vol. 137, nr. I, 1989, pp.
47 - 75.
No. 9 G. Dltaene and A.P. Barten, When it all began: the l936 Tinbergen model revisited, Economic Mode!ling, vol. 6, no. 2, 1989, pp. 203 - 219.
No. ]0 F. van der Ploeg and A.J. de Zeeuw, Conflict over arms accumulation in market and cormnand economies, in F. van der Ploeg and A.J. de Zeeuw (eds.), Dynmrric Policy Games in Econornics, Contributions to Economic Analysis 181, Amster- dam: Elsevier Science Publishers B.V. (North-Holland), 1989, pp. 91 - 119.
No. ll J. Driffill, Macroeconomic policy games with incomplete information: some extensions, in F. van der Ploeg and A.J. de Zeeuw (eds.), Dynarnic Poliry Cames in Economics, Contributions to Economic Analysis 181, Amsterdam: Elsevier Science Publishers B.V. (North-Holland), 1989, pp. 289 - 322.
No. l3 R.1.M. Alessie and A. Kapteyn, Consumption, savings and demography, in A. Wenig, K. F. Zimmermann (eds. ), Demographic ChangeatdEconomic Development, BerlinlHeidelberg: Springer-Verlag, 1989, pp. 272 - 305.
No. 14 A. Hoque, J.R. Magnus and B. Pesaran, The ezact multi-period mean-square forecast error for the first-order autoregressive model, Journa! of Econometrics, vol. 39, no. 3, 1988, pp. 327 - 346.
No. l5 R. Alessie, A. Kapteyn and B. Melenberg, The effects of liquidity constraints on consumption: estimation from household panel data, Europeatt Econonric Revie~v, vol. 33, no. 2~3, 1989, pp. 547 - 555.
No. 16 A. Holly and J.R. Magnus, A note on instrumental variables and mazimum
likeli-hood estimation procedures, Artnales d'Économie et de Staristique, no. (0, April-1une, 1988, pp. 121 - 138.
No. 17 P. ten Hacken, A. Kapteyn and I. Woittiez, Unemployment benefits and the labor market, a microlmacro approach, in B.A. Gustafsson and N. Anders Klevmarken (eds.), 77te Politica! Economy of Socia! Security, Contributions to Economic Analysis 179, Amsterdam: Elsevier Science Publishers B.V. (North-Holland), 1989, pp. l43 - 164.
No. 18 T. Wansbeek and A. Kapteyn, Estimation of the error-components model witli incomplete panels, lourna! of Eco~onterrics, vol. 41, no. 3, 1989, pp. 341 - 361. No. 19 A. Kapteyn, P. Kooreman and R. Willemse, Some methodological issues in the
implementation of subjective poverty definitions, 7he Journa! of Human Resources, voL 23, no. 2, 1988, PP. 222 - 242.
Th. van de Klundert and F. van der Plceg, Fiscal policy and finite lives in interdependent economies with real and nominal wage rigidity, Oxford Econornic Papers, vol. 41, no. 3, 1989, pp. 459 - 489.
J.R. Magnus and B. Pesaran, The exact multi-period mean-square forecast error for
the first-order autoregressive model with an intercept, lourna( of Economerrics, vol. 42, no. 2, 1989, pp. l57 - I79.
F. van der Ploeg, Two essays on politícal economy: (i) The political economy of overvaluation, 77te Econontic lournal, vol. 99, no. 397, 1989, pp. 850 - 855; (ii) Election outcomes and the stockmarket, Europeattlourrta!ojPolitica! Econorrry, vol. 5, no. I, 1989, pp. 21 - 30.
No. 23 I.R. Magnus and A.D. Woodland, On the maximum likelihood estimation of multivariate regression models containing serially correlated error components,
Irttentatiórtal Ecottornic Review, vol. 29, no. 4, 1988, pp. 707 - 725.
No. 24 A.J.J. Talman and Y. Yamamoto, A simplicial algorithm for stationary point problems on polytopes, Mathematics of Operations Research, vol. 14, no. 3, 1989, pp. 383 - 399.
No. 25 E. van Damme, Stable equilibria and forward induction, lournal ofEcortomic7lteory,
No. 26 A.P. Barten and L.1. Bettendorf, Price fonnation of fish: An application of an inverse demand system, European Economic Review, vol. 33, no. 8, 1989, pp. 1509 - 1525. No. 27 G. Noldeke and E. van Damme, Signalling in a dynamic labour market, Rer~iew oj
Econornic Studies, vol. S7 (1), no. 189, 1990, pp. 1- 23.
No. 28 P. Kop Jansen and Th. ten Raa, The choice of model in the construction of input-output coe(ficients matrices, Inlernationa! Ecortomic Review, vol. 31, no. 1, 1990, pp. 213 - 227.
No. 29 F. van der Ploeg and A.J. de Zeeuw, Perfect equilibrium in a model of cotnpetitive arms accumulation, Internationa! Ecatomic Review, vol. 31, no. 1, 1990, pp. l3l
-146.
No. 30 1.R. Magnus and A.D. Woodland, Separability and aggregation, Econontica, vol. S7,
no. 226, 1990, pp. 239 - 247.
No. 31 F. van der Ploeg, International interdependence and policy coordination in economies
with real and nominal wage rigidity, Greek Economic Review, vol. I0, no. 1, June 1988, pp. i - 48.
No. 32 E. van Damme, Signaling and forward induction in a market entry context, Operations Research Proceedings 1989, Berlin-Heidelberg: Springer-Verlag, 1990, pp. 4S - S9.
No. 33 A.P. Barten, Toward a levels version of the Rotterdam and related detnand systems, Coruributíorts to Operarions Research and Ecatomics, Cambridge: MIT Press, 1989, pp. 441 - 465.
No. 34 F. van der Ploeg, International coordination of monetary policies under alternative exchange-rate regimes, in F. van der Ploeg (ed.), Advanced L.ectures in Quantitative Economics, London-Orlando: Academic Press Ltd., 1990, pp. 91 - 121.
No. 35 Th. van de Klundert, On socioeconomic causes of 'wait unemployment', Eurapean Ecatonric Review, vol. 34, no. S, 1990, pp. 1011 - 1022.
No. 36 R.I.M. Alessie, A. Kapteyn, ].B. van Lochem and T.1. Wansbeek, lndividual effects in utility consistent models of demand, in 1. Hartog, G. Ridder and 1. Theeuwes (eds.), Pane! Data arrd Labor Market Studies, Amsterdam: Elsevier Science Publishers B.V. (North-Holland), 1990, pp. 2S3 - 278.
No.37 F. van der Ploeg, Capital accumulation, inftation and long-run contlict in international objectives, Oxford Ecorrontic Papers, vol. 42, no. 3, I990, pp. SO1
-525.
No. 38 Th. Nijman and F. Palm, Parameter identification in ARMA Processes in tlre
presence of regular but incomplete sampling, Jounra[ of lïme Series Anolysis, vol. l i, no. 3, I990, pp. 239 - 248.
No. 39 Th. van de Klundert, Wage differentials and employment in a two-sector model with
No.40 Th. Nijman and M.F.J. Steel, Exclusion restrictions in instrumental variables
equations, Econometric Reviews, vol. 9, no. 1, 1990, pp. 37 - 55.
No. 4l A. van Soest, I. Woittiez and A. Kapteyn, Labor supply, income taxes, and hours restrictions in the Netherlands, Jourrra( of Hrtman Resottrces, voL 25, no. 3, 1990, pp. 5l7 - 558.
No. 42 Th.C.M.J. van de Klundert and A.B.T.M. van Schaik, Unetnployment persistence
and loss of productive capacity: a Keynesian approach, Journal of Macro- economics,
vol. I2, no. 3, 1990, pp. 363 - 380.
No. 43 Th. Nijman and M. Verbeek, Estimation of lime-dependent parameters in linear
models using cross-sections, panels, or both, Jounta! of Econontetrics, vol. 46, no.
3, 1990, pp. 333 - 346.
No. 44 E. van Damme, R. Selten and E. Winter, Alternating bid bargaining with a smallest money unit, Games and Econorrtic Belaavior, vol. 2, no. 2, 1990, pp. 188 - 201. No. 45 C. Dang, The D~-triangulation of R" for simplicial algorithms for computing solutions
of nonlinear equations, Mathenwtics of Operations Research, vol. 16, no. I, 1991, pp. 148 - 161.
No. 46 Th. Nijman and F. Palm, Predictive accuracy gain from disaggregate sampling in ARIMA models, lourna! of Business ác Econarnic Statistics, vol. 8, no. 4, 1990, pp. 405 - 415.
No. 47 J.R. Magnus, On certain moments relating to ratios of quadratic fomis in normal variables: further results, Sankhya: 7he lndian Journal ojStatistics, vol. 52, series B, part. I, I990, PP. 1- 13.
No. 48 M.F.J. Steel, A Bayesian analysis of simultaneous equation models by combining
recursive analytical and numerical approaches, Journa! of Econanetrics, vol. 48, no.
112, 1991, pp. 83 - 1 l7.
No. 49 F. van der Plceg and C. Withagen, Pollution control and the ramsey problem,
Ettvironntental and Resource Ecatomics, vol. 1, no. 2, 1991, pp. 2I5 - 236.
No. 50 F. van der Ploeg, Money and capital in interdependent economies with overlapping
generations, Econoneica, vol. 58, no. 230, 1991, pp. 233 - 256.
No. SI A. Kapteyn and A. de Zeeuw, Changing incentives for economic research in the Netherlands, European Eca~omic Review, vol. 35, no. 213, 1991, pp. 603 - 611. No. 52 C.G. de Vries, On the relation between GARCH and stable processes, Journa! of
Econontetrics, vol. 48, no. 3, 1991, pp. 3l3 - 324.
No. 53 R. Alessie and A. Kapteyn, Habit formation, interdependent preCerences and
demograpliic ef(ects in the almost ideal demand system, The Economic Journal, vol.
I O l, no. 406, 1991, pp. 404 - 419.
No. 54 W. van Groenendaal and A. de Zeeuw, Control, coordination and con(lict on
No. 55 F. van der Ploeg and A.J. Markink, Dynamic policy in linear models with ratioual expectations of future events: A computer package, Computer Science in Economics and Management, vol. 4, no. 3, 1991, pp. 175 - 199.
No. 56 H.A. Keuzenkamp and F. van der Plceg, Savings, investment, government finance, and the current account: The Dutch experience, in G. Alogoskoufis, L. Papadetnos and R. Portes (eds.), External Constraints on Macrceconomic Policy: The European Experience, Cambridge: Cambridge University Press, 1991, pp. 219 - 263.
No. 57 Th. Nijman, M. Verbeek and A. van Soest, The efficiency of rotating-panel designs in an analysis-of-variance model, Journal of Econometrics, vol. 49, no. 3, 1991, pp.
373 - 399.
No. 58 M.F.J. Steel attd J.-F. Richard, Bayesian multivariate exogeneity analysis - an applícation to a UK money demand equation, Journal of Econometrics, vol. 49, no.
112, 1991, PP. 239 - 274.
No. 59 Th. Nijman and F. Palm, Generalized least squares estimation of linear tnodels containing rational future expectations, lnternational Economic Review, vol. 32, no. 2, 1991, pp. 383 - 389.
No. 60 E. van Damme, Equilibrium selection in 2 x 2 games, Revista Espanola de
Economia, vol. 8, no. I, 1991, pp. 37 - 52.
No. 61 E. Bennett and E. van Damme, Demand commitment bargaining: the case of apex games, in R. Selten (ed.), Game Equilibrium Models li1 - Strategic Bargaining, Berlin: Springer-Verlag, 1991, pp. I 18 - 140.
No.62 W. Giith and E. van Damme, Gorby games - a game theoretic analysis of disarmament campaigns and the defense e(ficiency - hypothesis -, in R. Avenhaus, H. Karkar and M. Rudnianski (eds.), Defense Decision Making - Analytical Support and Crisis Management, Berlin: Springer-Verlag, 1991, pp. 215 - 240.
No. 63 A. Roell, Dual-capacity trading and the quality of the market, Journa[ ojFinancia! I~:rern~ediarion, vol. 1, no. 2, 1990, pp. 105 - 124.
No. 64 Y. Dai, G. van der Laan, A.1.1. Talman and Y. Yamamoto, A simplicial algorith~n for the nonlinear stationary point problem on an unbounded polyhedron, Siant Journol ojOprimiZarion, voL 1, no. 2, 1991, pp. 15l - 165.
No.65 M. McAleer and C.R. McKenzie, Keynesian and new classical models of unemployment revisited, The Eco~tomic Journal, vol. 101, no. 406, 1991, pp. 359 - 381.
No. 66 A.J.J. Talman, General equilibrium programming, NiettwArchiejvoor Wiskunde, vol. 8, no. 3, 1990, pp. 387 - 397.
No. 67 J.R. Magnus and B. Pesaran, The bias of forecasts from a 6rst-order autoregression,
No. 68 F. van der Plceg, Macroeconotnic policy coordination issues during the various phases of economic and monetary integration in Europe, Etrropemt Economy - The Economics ojEMU, Commission of the European Communities, special edition no.
1, 1991, PP. 136 - 164.
No. 69 H. Keuzenkamp, A precursor to Muth: Tinbergen's 1932 model of rational
expectations, The Econornic Journal, vol. lO1, no. 408, 199I, pp. l245 - 1253.
No. 70 L. Zou, The target-incentive system vs. the price-incentive system under adverse
selection and the ratchet effect, Jotrrnal ojPublic Economics, vol. 46, no. 1, 1991, PP. 51 - 89.
No. 71 E. Bomhoff, Between price reform and privatization: Eastern Europe in transition, Finatzmarkt und Portfolio Managentent, vol. 5, no. 3, 1991, pp. 241 - 251. No. 72 E. Bomhoff, Stability of velocity in the major industrial countries: a Kalman filter
approach, Intentational Monetary Fwtd Staff Papers, vol. 38, no. 3, 1991, pp. 626
- 642.
No. 73 E. Bomhoff, Currency convertibility: when and how? A contribution to tlte Bulgarian debate, Kredit ur:d Kapital, vol. 24, no. 3, 1991, pp. 4l2 - 431.
No. 74 H. Keuzenkamp and F. van der Plceg, Perceived constraints for Dutch unemployment policy, in C. de Neubourg (ed.), 77te Art ojFull Employment - Unemplaymeru Policy in Open Ecwtanies, Contributions lo Economic Analysis 203, Amsterdam: Elsevier Science Publishers B.V. (North-Holland), 1991, pp. 7- 37.
No. 75 H. Peters and E. van Damme, Characterizing the Nash and Raif(a bargaining solutions by disagreement point axions, Mathematics of OperarioruResearclt, vol. 16, no. 3, 1991, pp. 447 - 461.
No. 76 P.1. Deschamps, On the estimated variances of regression ccefficients in misspecified error components models, Econometric 7lteory, vol. 7, no. 3, 1991, pp. 369 - 384. No. 77 A. de Zeeuw, Note on ' Nash and Stackelberg solutions in a differential game motJel of capitalism', Journal ofEconomic Dyttamics and Control, vol. 16, no. l, 1992, pp. 139 - 145.
No. 78 J.R. Magnus, On the fundamental bordered matrix of linear estimation, in F. van der Plceg (ed.), Advartced L.ectures in Quantitative Econanics, London-0rlando: Academic Press Ltd., 1990, pp. 583 - 604.
No. 79 F. van der Plceg and A. de Zeeuw, A differential game of international pollution
control, Systenu and Contro! Letrers, vol. l7, no. 6, 1991, pp. 409 - 414.
No. 80 Th. Nijman and M. Verbeek, The optimal choice of controls and pre-experimen- tal observations, Jottntal ojEcor:ometrics, vol. 51, no. 112, 1992, pp. 183 - 189.
No. 81 M. Verbeek and Th. Nijman, Can cohort data be treated as genuine panel data?,
No. 82 E. van Damme and W. Guth, Equilibrium selection in the Spence signaling game, in
R. Selten (ed.), Game EquilibriurrrMode[s II - Metlrods, Mora(s, andMarkets, Berlin:
Springer-Verlag, 1991, pp. 263 - 288.
No. 83 R.P. Gilles and P.H.M. Ruys, Characterization of economic agents in arbitrary communication structures, Nieuw Archief voor Wiskurtde, vol. 8, no. 3, 1990, pp. 325 - 345.
No. 84 A. de Zeeuw and F. van der Ploeg, Difference games and policy evaluation: a conceptual framework, Oxford Ecortorrtic Papers, vol. 43, no. 4, 1991, pp. 612
-636.
No. 85 E. van Danune, Fair division under asymmetric information, in R. Selten (ed.),
Rariona! Interactiar - Essays in Honor of John C. Harsanyi, BerlinlHeidelberg:
Springer-Verlag, 1992, pp. I21 - 144.
No. 86 F. de Jong, A. Ketnna and T. Kloek, A contribution to event study methodology with an application to the Dutch stock market, Jountal of Banking and Finance, vol. 16, no. 1, l992, pp. 1 I- 36.
No. 87 A.P. Barten, The estimation of mixed demand systems, in R. Bewley and T. Van Hoa (eds.), Caruributions to Coruwtter Dernand artd Econometrics, Essays in Hatour
of Henri 7hei[, Basingstoke: The Macmillan Press Ltd., 1992, pp. 31 - 57. No. 88 T. Wansbeek and A. Kapteyn, Simple estimators for dynamic panel data models with
errors in variables, in R. Bewley and T. Van Hoa (eds.), Corrtributions to Corrstuner Dentand and Ecatolnetrics, Essays in Honour of Henri 77tei(, Basingstoke: The
Macmillán Press Ltd., 1992, pp. 238 - 251.
No. 89 S. Chib, 1. Osiewalski and M. Steel, Posterior inference on the degrees of freedom
parameter in multivariate-t regression models, Econorrtics Letters, vol. 37, no. 4,
1991, pp. 391 - 397.
No. 90 H. Peters and P. Wakker, lndependence of irrelevant alternatives and revealed group preferences, Ecormmetrica, vol. 59, no. 6, 1991, pp. 1787 - 1801.
No. 91 G. Alogoskoufis and F. van der Ploeg, On budgetary policies, growth, and external
deficits in an interdependent world, Journal of the Japmrese and buernariortal
Econonties, vol. 5, no. 4, 1991, pp. 305 - 324.
No. 92 R.P. Gilles, G. Owen and R. van den Brink, Games with permission structures: The conjunctive approach, Internationa! Journa!of Gante Theory, vol. 20, no. 3, 1992, PP~ 277 - 293.
No. 93 J.A.M. Potters, I.J. Curiel and S.H. Tijs, Traveling salesman games, Matltentotica! Progranvning, vol. 53, no. 2, 1992, pp. 199 - 2l 1.
No. 95 A. van den Nouweland, P. Borm and S. Tijs, Allocation rules for hypergrapli
communication situations, Internatíonal Journal oJ Game 77teory, vol. 20, no. 3,
1992, PP. 255 - 268.
No. 96 E.J. Bomhoff, Monetary reform in Eastern Europe, European Economic Review, vol. 36, no. 213, 1992, pp. 454 - 458.
No. 97 F. van der Ploeg and A. de Zeeuw, International aspects of pollution control, Ern~ironmental and Resource Economics, vol. 2, no. 2, 1992, pp. 117 - 139. No. 98 P.E.M. Borm and S.H. Tijs, Strategic claim games corresponding to an NTU-game,
Gmnes artd Econonric Behavior, vol. 4, no. l, 1992, pp. 58 - 7l.
No. 99 A. van Scest and P. Kooreman, Coherency of the indirect translog demand system
with binding nonnegativity constraints, Jaurnal oj Ecwtometrics, vol. 44, no. 3,
1990, pp. 391 - 400.
No. 100 Th. ten Raa and E.N. Wolff, Secondary products and the measurement of productivity growth, Regional Science and Urbm: Economics, voL 2l, no. 4, 1991, pp. 581 - 615.
No. 10l P. Kooreman and A. Kapteyn, On the empirical implementation of some game theoretic models of household labor supply, 7T:e Jortnta! ojHurnan Resources, vol. 25, no. 4, 1990, pp. 584 - 598.
No. 102 H. Bester, Bertrand equilibrium in a differentiated duopoly, huernationa! Ecortonric Revierv, vol. 33, no. 2, 1992, pp. 433 - 448.
No. 103 1.A.M. Potters and S.H. Tijs, The nucleolus of a matrix game and other nucleoli, Mathentatics of t7pera[ions Research, vol. l7, no. 1, I992, pp. 164 - 174. No. 104 A. Kapteyn, P. Kooreman and A. van Soest, Quantity rationing and concavity in a
flezible household labor supply model, Review of Economics and Statistics, vol. 72, no. I, 1990, pp. 55 - 62.
No. l05 A. Kapteyn and P. Kooreman, Household labor supply: What kind of data can tell us how many decision makers there are?, European Ecortornic Review, vol. 36, no. 213, 1992, pp. 365 - 371.
No. 106 Th. van de Klundert and S. Smulders, Reconstructing growth theory: A survey, De Ecortonrisr, vol. l40, no. 2, 1992, pp. l77 - 203.
No. 107 N. Rankin, [mperfect competition, ezpectations and tlte multiple effects of monetary growth, The Ecorrontic Journal, vol. 102, no. 413, 1992, pp. 743 - 753. No. 108 ]. Greenberg, On the sensitivity of von Neumann and Morgenstern abstract stable
sets: The stable and the individual stable bargaining set, International Jounral oJ Game 7heory, vol. 21, no. 1, 1992, pp. 41 - 55.
No. 110 M. Verbeek and Th. Nijman, Testing for selectivity bias in panel data models, InrernatronalEconanic Ret7e~v, vol. 33, no. 3, 1992, pp. 681 - 703.
No. 1 I 1 Th. Nijman and M. Verbeek, Nonresponse in panel data: The impact on estimates of a life cycle consumption function, Journa! of Applied Ecarometrics, vol. 7, no. 3, 1992, pp. 243 - 257.
No. I 12 I. Bomze and E. van Damme, A dynamical characterization of evolutionarily stable states, Ararals of Operarions Research, vol. 37, 1992, pp. 229 - 244.
No. 113 P.1. Deschatnps, Expectations and intertempora) separability in an empirical model of consumption and investment under uncertainty, Empirical Econornics, vol. 17, no. 3, 1992, pp. 419 - 450.
No. 114 K. Kamiya and D. Talman, Simplicial algoritlun for computing a core eletnent itt a balanced game, Jountal of rlte Operations Research, vol. 34, no. 2, 1991, pp. 222
-228.
No. 115 G. W. imbens, An efficient method of moments estimator for discrete choice models with choice-based sampling, Econanerrica, vol. 60, no. 5, 1992, pp. I187 -1214. No. ! 16 P. Borm, On perfectness concepts for bimatrix games, OR Spektrum, vol. 14, no.
1, 1992, pp. 33 - 42.
No. 117 A.P. Jurg, 1. Garcia Jurado and P.E.M. Borm, On modifications of the concepts of perfect and proper equilibria, OR Spektrum, vol. 14, no. 2, 1992, pp. 85 - 90. No. l 18 P. Borm, H. Keiding, R.P. McC.ean, S. Oortwijn and S. Tijs, The compromise value
for NTU-games, hNernatioaralJorarna! oj Game 7heary, vol. 21, no. 2, 1992, pp. 175 - 189.
No. 119 M. Maschler, J.A.M. Potters and S.H. Tijs, The general nucleolus and the reduced game property, lruerrtatiortal Journal of Cante 77teory, vol. 2l, no. 1, 1992, pp. 85
-l06.
No. 120 K. Wárneryd, Communication, correlation and symmetry in bargaining, Econonrics Letters, vol. 39, no. 3, 1992, pp. 295 - 300.
No. 121 M.R. Baye, D. Kovettock and C.G. de Vries, It takes two to tango: equilibria in a model of sales, Carrtes and Ecorranic Behavior, vol. 4, no. 4, 1992, pp. 493 - 510. No. 122 M. Verbeek, Pseudo panel data, in L. Mátyás and P. Sevestre ( eds.), The EconanetricsofPanelData, Dordrecht: KluwerAcademic Publishers, 1992, pp. 303 - 315.
No. l23 S. van Wijnbergen, Intertemporal speculation, shortages and the political economy of price reform, T7:e Econontic Journal, vol. 102, no. 415, 1992, pp. 1395 - 1406. No. 124 M. Verbeek and Th. Nijman, Incomplete panels and selection bias, in L. Mátyás and
No. 125 7.1. Sijben, Monetary policy in a game-theoretic framework, Jahrbr7cher fiir Natiorralókorromie ~nd Statistik, vol. 210, no. 314, 1992, pp. 233 - 253.
No. 126 H.A.A. Verbon and M.J.M. Verhoeven, Decision making on pension schemes under
rational expectations, Jortrnal of Ecaromics, vol. 56, no. 1, 1992, pp. 71 - 97.
No. l27 L. Zou, Ownership structure and efficiency: An incentive mechanism approach, Jairna! oj Cornparative Ecatomics, vol. 16, no. 3, 1993, pp. 399 - 431.
No. l28 C. Fershtman and A. de Zeeuw, Capital accumulation and entry deterrence: A clarifying note, in G. Feichtinger (ed.), Dynamic Ecorromic Modefs mtd Optimal Contro(, Amsterdam: Elsevier Science Publishers B.V. (North-Holland), 1992, pp. 281 - 296.
No. 129 L. Bovenberg and C. Petersen, Public debt and pension policy, Fisca[Studies, vol. l3, no. 3, 1992, pp. 1- 14.
No. 130 R. Gradus and A. de Zeeuw, An employment game between government and firms, Optima! Coturol Applicatioru cF Methods, vol. 13, no. 1, 1992, pp. 55 - 71. No. 131 Th. Nijman and R. Beetsma, Empirical tests of a simple pricing model for sugar
futures, Arrrrales d'Écononrie et de Statistique, no. 24, 1991, pp. 12l - 131. No. 132 F. Groot, C. Withagen and A. de Zeeuw, Note on tlie open-loop Von Stackelberg
equilibrium in the Cartel versus Fringe model, 7Tte Ecoreomic Joun:al, vol. 102, no. 415, 1992, PP. 1478 - 1484.
No. 133 S. Eijffinger and N. Gruijters, On the effectiveness of daily intervention by the Deutsche Bundesbank and the Federal Reserve System in the U5 dollar - deutsche mark exchange market, in BaltenspergerlSinn (eds), Fxcltange-Rate Regimes and Currency Uniorts, Basingstoke: The Macmillan Press Ltd., 1992, pp. 131 - 156. No. 135 A. K. Bera and S. Lee, [nformation matrix test, parameter heterogeneity and ARCH:
a synthesis, Revietv of Ecortonric Studies, 60, 1993, pp. 229 - 240.
No. 136 H. G. Bloemen and A. Kapteyn, The joint estimation of a non-linear labour supply function and a wage equation using simulated response probabilities, Arrnoles d'Économie et de Statistique, No. 29, 1993, pp. 175 - 205.
No. 137 H. Bester, Bargaining versus price competition in markets with quality uncertainty, 7Tte Anrericon Economic Review, Vol. 83, No. 1, March 1993, pp. 278 - 288. No. 138 K. W~rneryd, Anarchy, uncertainty, and the emergence of property rights, Ecortottrics
and Politics, Vol. 5, No. l, March 1993, pp. 1- 14.
No. 139 A. L. Bovenberg and L.H. Goulder, Promoting investment under international capital
mobility: an intertemporal general equilibrium analysis, 77te Scmtdinavlan Jounta! of
Ecottomics, Vol. 95, No. 2, 1993, pp. 133 - I56.
No. l40 S. Eijffinger and E. Schaling, Central bank independence in twelve industrial
countries, Bmrca Naziortale de( Lavoro Quarterly Review, No. 184, March 1993, pp.
No. 14l S. Eijffinger and A. van Rixtel, The Japanese financial system and monetary policy: a descrip[ive review, Japan and rhe World Econorny, Vol. 4, No. 4, 1992, pp 291-309.
No. 142 A. L. Bovenberg, lnvestment-promoting policies in open economies: the importance of intergenerational and international distributional effects, Journa! of Public Econonucs, Vol. 51, 1993, North Holland, pp. 3-54 .
No. 143 A. 0zcam, G. Judge, A Bera and T. Yancey, The risk properties of a pre-test estimator for Zellner's seemingly unrelated regression model, Journalof Qumuirarive Ecortontics, Vol. 9, No. 1, January 1993, pp. 41-52.
No. l44 F. C. Drost and T. E. Nijman, Temporal aggregation of garch processes, Econometrica, Vol. 6l, No. 4, luly 1993, pp. 909-927.
No. 145 1. J. G. Lemtnen and S.C.W. Eijffinger, The degree of financial integration in the European Community, De Econontisr, Vol. 141, No. 2, 1993, pp. 189-213. No. 146 R. Sarin and P. Wakker, A simple axiomatization of nonadditive expected utility,
Econornerrica, Vol. 60, No. 6, November 1992, pp. 1255-1272.
No. 147 S. Muto, On licensing policies in bertrand competition, Games and Economic
Behaviour, 5, 1993, pp. 257-267.
No. 148 M. Verbeek and T. Nijman, Minimum MSE estimation of a regression model with fixed effects from a series of cross-sections, lournal of Economerrics, 59, 1993, pp. 125-136.
No. 149 R. de Groof and M. van Tuijl, Financial integration and fiscal policy in interdependent two-sector economies with real and nominal wage rigidity, Europemr Journa! of Political Econonry, Vol. 9, 1993, Nonh Holland, pp. 209-232. No. 150 A. van Soest, A. Kapteyn and P. Kooreman, Coherency and regularity of demand
systems with equality and inequality constraints, Jounta! of Econontetrics, Vol. 57, 1993, North- Holland, pp. 161-188.