LIMIT THEORY FOR EXPLOSIVELY COINTEGRATED SYSTEMS. PETER C. B. PHILLIPS and TASSOS MAGDALINOS COWLES FOUNDATION PAPER NO. 1244

Size: px
Start display at page:

Download "LIMIT THEORY FOR EXPLOSIVELY COINTEGRATED SYSTEMS. PETER C. B. PHILLIPS and TASSOS MAGDALINOS COWLES FOUNDATION PAPER NO. 1244"

Transcription

1 LIMIT THEORY FOR EXPLOSIVELY COINTEGRATED SYSTEMS BY PETER C. B. PHILLIPS ad TASSOS MAGDALINOS COWLES FOUNDATION PAPER NO COWLES FOUNDATION FOR RESEARCH IN ECONOMICS YALE UNIVERSITY Box New Have, Coecticut

2 Ecoometric Theory, 24, 2008, Prited i the Uited States of America+ doi: s LIMIT THEORY FOR EXPLOSIVELY COINTEGRATED SYSTEMS PETER C.B. PHILLIPS Cowles Foudatio for Research i Ecoomics, Yale Uiversity Uiversity of Aucklad ad Uiversity of York TASSOS MAGDALINOS Uiversity of Nottigham A limit theory is developed for multivariate regressio i a explosive coitegrated system+ The asymptotic behavior of the least squares estimator of the coitegratig coefficiets is foud to deped upo the precise relatioship betwee the explosive regressors+ Whe the eigevalues of the autoregressive matrix Q are distict, the cetered least squares estimator has a expoetial Q rate of covergece ad a mixed ormal limit distributio+ No cetral limit theory is applicable here, ad Gaussia iovatios are assumed+ O the other had, whe some regressors exhibit commo explosive behavior, a differet mixed ormal limitig distributio is derived with rate of covergece reduced to M+ I the latter case, mixed ormality applies without ay distributioal assumptios o the iovatio errors by virtue of a Lideberg type cetral limit theorem+ Covetioal statistical iferece procedures are valid i this case, the statioary covergece rate domiatig the behavior of the least squares estimator+ 1. INTRODUCTION Autoregressios with a explosive root 6u6 1 came to promiece after the early work of White ~1958! ad Aderso ~1959!+ Assumig Gaussia iovatio errors, these authors derived a Cauchy limit theory for the cetered least squares estimator with rate of covergece u + The theory was geeralized by Mijheer ~2002! to o-gaussia explosive processes geerated by iovatios satisfyig a stability property+ I each of these works, o cetral limit theory applies, ad the asymptotic distributio of the least squares estimator is characterized by the distributioal assumptios imposed o the iovatios+ Phillips thaks the NSF for research support uder grat SES Address correspodece to Peter C+B+ Phillips, Departmet of Ecoomics, Yale Uiversity, P+O+ Box , New Have, CT , USA; peter+phillips@yale+edu Cambridge Uiversity Press $

3 Z Z 866 PETER C.B. PHILLIPS AND TASSOS MAGDALINOS I this paper, we cosider a explosively coitegrated system y t Ax t «t, x t Qx u t, Q I K C, C diag~c 1,+++,c K!, c i ~,2! ~0,! i, (1) (2) where A is a m K matrix of coitegratig coefficiets, x t is a K-vector of explosive autoregressios iitialized at x 0 0, ad v t ~«t, u t! is a sequece of idepedet, idetically distributed ~0, S! radom vectors with absolutely cotiuous desity, where S is a positive defiite matrix partitioed coformably with v t as S diag~s ««, S uu!+ We deote by u i 1 c i the ith diagoal elemet of Q ad by 7Q7 max 1iK 6u i 6 the spectral orm of Q+ The asymptotic behavior of the least squares estimator A y t x t 1 x t x t is foud to deped o the relatioship betwee the regressors i ~2!, i+e+, o the precise form of the matrix Q+ As Theorem 2+1, which follows, shows, the rak of the limit matrix of the ormalized sample secod momets, ad hece the order of magitude of ~ x t x t! 1, is determied exclusively by Q+ Whe Q yields a osigular limit i Theorem 2+1, AZ A is foud to have a Q rate of covergece ad a mixed ormal limitig distributio, uder the assumptio of Gaussia iovatios ~cf+ Aderso, 1959!+ But whe the limit momet matrix of Theorem 2+1 is sigular, A A has a degeerate mixed ormal limitig distributio with covergece rate reduced to The asymptotics i the sigular case are obtaied by rotatig the regressio coordiates i a way that the sigularity is elimiated ad cetral limit theory applies+ Cosequetly, the mixed ormal limit theory i the sigular case applies without ay distributioal assumptios o the iovatio errors+ Explosive systems are useful i modelig periods of extreme behavior i ecoomic ad fiacial variables+ Ecoomic growth amog the Asia dragos durig the 1980s ad recet growth i Chia provide examples of mildly explosive growth i macroecoomic variables+ Hyperiflatio i Germay i the 1920s ad Yugoslavia i the 1990s are examples of some of the may historical istaces of explosive behavior i prices+ Fiacial bubbles i asset prices are aother example, the recet rise ad subsequet fall i price of Iteret stocks i the NASDAQ market creatig ad destroyig some $8 trillio of shareholder wealth+ To the extet that periods of explosive movemet i such variables ifluece ecoomic decisios or cotamiate other variables, we may expect models of explosive coitegratio such as ~1! ad ~2! to be relevat i relatig these variables+ Whe there is a sigle source of the extreme movemet, the such a system may also have explosively coitegrated regressors, ad the degeeracy described earlier may occur+

4 Z 2. RESULTS LIMIT THEORY FOR EXPLOSIVELY COINTEGRATED SYSTEMS 867 We develop a limit theory for the cetered least squares estimator A A «t x t 1 x t x t + It turs out that the asymptotic order of AZ A depeds o the rak of the limit of the ormalized sample momet matrix x t x t + The latter ca be derived, usig a similar method to Aderso ~1959!, i terms of the radom vector X Q Q j u j, j1 (3) where the series coverges almost surely by virtue of the martigale covergece theorem+ THEOREM 2+1+ The sample momet matrix of the explosive process (2) satisfies Q x t x t Q r a+s+ Q j X Q X Q Q j as r, (4) j0 where X Q is the radom vector defied i (3). Note that the almost sure limit j0 Q j X Q X Q Q j of the ormalized sample momet matrix is ot always osigular+ Deote the ith elemet of the radom vector X Q by X! Q ad defie the ~i matrices M Q : u i u j u i u j 1 : i, j $1,+++,K% ad X Q : diag~x Q ~1!,+++,X Q ~K!!+ Because u 1 admits a absolutely cotiuous desity, X Q ~i! 0 almost surely for each i+ Thus, the idetity j0 Q j X Q X Q Q j X Q M Q X Q implies that j0 Q j X Q X Q Q j is osigular wheever the matrix M Q is osigular, i+e+, if ad oly if c i c j for all i j ~cf+ Lemma 4+3!+ O the other had, whe ay two localizig coefficiets c i, c j are the same, the matrix M Q will have two idetical colums ad will, therefore, be sigular+ We begi by discussig the osigular asymptotic momet matrix case+

5 868 PETER C.B. PHILLIPS AND TASSOS MAGDALINOS THEOREM 2+2+ For the explosive coitegrated system geerated by (1) ad (2) with v t d N~0, S! ad c i c j for all i j, the followig limit theory applies as r : vec@~ AZ A!Q # MN0, j 1 Q j X Q X Q Q S ««. Remarks 2.1. j0 ~i! The assumptio of Gaussia iovatios is essetial to obtai a mixed ormal limitig distributio for the least squares estimator+ This is because, despite beig asymptotically equivalet to a martigale array ~see ~34! i Sect+ 4!, the sample covariace does ot satisfy the requiremet of uiform asymptotic egligibility or the Lideberg coditio ~cf+ Hall ad Heyde, 1980, Sect+ 3+2!+ As a result, o cetral limit theory applies i geeral, ad mixed ormality requires Gaussia iovatios, as i the AR~1! case of Aderso ~1959!+ ~ii! Whe v t d N~0, S!, X Q d N~0, j1 Q j S uu Q j!+ ~iii! I the simplest case of a two-equatio system, K 1, ad so x t, A a, ad Q u are scalar+ Lettig Z be a N~0,1! variate, the previous remark yields ~u 2 1! 102 X Q d N~0, S uu! d S 102 uu Z, u Q j X Q X Q Q j 2 u 2j X 2 Q d j0 j0 ~u 2 1! S 2 uu Z 2 + Thus, Theorem 2+2 reduces to u ~ a[ a! MN0, S ««~u 2 1! S uu 2 Z 2 u 2 where Y ad Z are idepedet N~0,1! variates, or u S uu d u 2 1 S ««u S uu 102 Y Z, u 2 1 ~ a[ a! S ««C, where C is a stadard Cauchy variate+ I the geeral case, the exact form of the limitig distributio of Theorem 2+2 ca be obtaied by usig a matrix quotiet argumet, as i Phillips ~1985!+ We ow tur to the discussio of the limit theory i the case of two or more equal localizig coefficiets+ We have see that this case gives rise to a sigular limit matrix for the sample variace, reflectig the fact that the regressors x t are themselves explosively coitegrated+ Because the mixig radom matrix j0 Q j X Q X Q Q j is sigular, the limit theory of Theorem 2+2 does ot apply+ The asymptotic behavior of the least squares estimator ca be determied by a rotatio of coordiates to isolate the explosive ad oexplosive behavior, a

6 LIMIT THEORY FOR EXPLOSIVELY COINTEGRATED SYSTEMS 869 method used by Park ad Phillips ~1988, 1989! i the settig of coitegrated processes+ Here, however, the rotatio is radom ad is determied by the limit vector X Q + We start by groupig together the repeated diagoal elemets of Q+ This ca be doe without loss of geerality by premultiplyig ~2! by a appropriate permutatio matrix ~i+e+, a square matrix cosistig of zeros ad oes that cotais exactly oe elemet 1 i each row ad each colum!+ If there are p groups of repeated diagoal elemets of Q the autoregressive matrix ca be rearraged as F diag~f 1, F 2!, F 1 diag~u 1 I r1,+++,u p I rp!, F 2 diag~w 1,+++,w Kr! p r r i, (5) i1 where all w i ad u i are diagoal elemets of Q with w s u l for all s, l ad w i w j, u i u j for all i j+ This effectively rearrages the system of equatios i ~2! ito a system of the form x1t J t u1 x1 u1t J J x pt u p x p x p1, F 2 x p1, (6) u pt u p1, t, where x it R r i icludes the regressors i ~2! that cotai the repeated root u i for each i $1,+++,p% ad x p1, t R Kr icludes the regressors that cotai all distict diagoal elemets of Q+ Lettig xi t ~x 1t,+++,x pt, x p1, t! ad ui t ~u 1t,+++,u pt, u p1, t!, ~6! ca be obtaied from ~2! as follows+ Cosider the K K permutatio matrix P that trasforms x t ito xi t : Px t xi t + The, usig orthogoality of permutatio matrices, ~2! yields xi t PQx ui t PQP Px ui t FxI ui t, (7) where F PQP has the explicit form give i ~5! ad, by orthogoality of P, satisfies the useful idetity F j PQ j P for all j N+ (8) Similarly, we ca write ~1! i terms of xi t as y t AP xi t «t CxI t «t, (9) where C AP + Because ZC C ~ AZ A!P, the asymptotic behavior of AZ is completely determied by that of ZC + I what follows, we show that oly the

7 870 PETER C.B. PHILLIPS AND TASSOS MAGDALINOS first r rows of the permutatio matrix P will cotribute to the limitig distributio of M ~ AZ A!+ It is therefore coveiet to partitio P as P 1 rk P P 2 ~Kr!K, where, by the orthogoality of P, P 1 ad P 2 satisfy P 1 P 1 I r, P 2 P 2 I Kr, P 1 P 2 0, P 1 P 1 P 2 P 2 I K + (10) I particular, the first two equalities i ~10! imply that rak~p 1! r ad rak~p 2! K r+ Coformably, we partitio F j as F j diag~f 1 j, F 2 j!+ The partitioed form of P together with ~8! the gives rise to the idetities F 1 j P 1 Q j P 1, F 2 j P 2 Q j P 2, P 1 Q j P 2 0, (11) Q j P 1 F 1 j P 1 P 2 F 2 j P 2 (12) for all j N+ The limit theory for the coitegrated system ~9! ad ~7! is derived by rotatig the regressio space i a directio orthogoal to X F : P 1 X Q ~P 1 Q j P 1!P 1 u j F j 1 P 1 u j, j1 j1 where the last equality is obtaied usig ~11!+ Correspodig to the partitio of P 1 x t, defie X F ~X F1,+++,X Fp!, X Fi R r i, ad HFi X Fi 0~X Fi X Fi! 102 for each i $1,+++,p%+ We cosider a r i ~r i 1! orthogoal complemet H 4i to each H Fi satisfyig H 4i H Fi 0 ad H 4i H 4i I ri 1 almost surely for all i $1,+++,p%+ The H H 41 0 J H 42 J 0 0 J J J J J 0 0 J H 4p 0 H F1 0 J H F2 J 0 0 J J J J J 0 0 J H Fp J 0 I Kr (13)

8 I is a K K orthogoal matrix that ca be partitioed as H 4 HF 0 4 H U U, U 4 F ~Kr!~rp!, U F 0 r~kr! 0 ~Kr!p I Kr H 4 diag~h 41,+++,H 4p!, H F diag~h F1,+++,H Fp!+ (14) By costructio, the orthogoal complemet matrix H 4 satisfies H 4 X F H 4 H F 0 ad H 4 H 4 I rp almost surely+ Although H 4 is ot uique, its outer product is uiquely defied by the relatio H 4 H 4 I r H F H F a+s+ (15) ~see, e+g+, Abadir ad Magus, 2005, 8+67!+ Moreover, ~14! implies a similar relatioship betwee U F ad U 4, amely, U 4 U F 0, U 4 U 4 I rp ad U 4 U 4 I K U F U F a+s+ (16) ad Applyig the orthogoal trasformatio H to the explosive regressor yields z t HxI 1t, z 2t #, z 1t H 4 P 1 x t R rp, z 2t U F Px t R Krp, (17) ZC C LIMIT THEORY FOR EXPLOSIVELY COINTEGRATED SYSTEMS 871 «t xi t H H x t I «t z t z t z t 1 x t H With this rotatio, the limit matrices of both 1 H, H+ (18) z t z t ad ~ z t z t! 1 are well defied after appropriate ormalizatio+ To see this, first observe that, i view of the idetities ~H 4 F 1 1 H 4! j H 4 F 1 j H 4, H 4 F 1 i H F 0 j N, i Z, (19) z 1t satisfies the reverse autoregressio z 1t ~H 4 F 1 1 H 4!z 1 H 4 F 1 1 P 1 u, (20) which, upo recursio, yields for each t t z 1t ~H 4 F 1 1 H 4! t z 1 H 4 F j 1 P 1 u tj + (21) j1

9 872 PETER C.B. PHILLIPS AND TASSOS MAGDALINOS Proofs of ~19! ad ~21! are give i Sectio 4+ Usig ~10! ad ~12! the secod term of ~21! ca be writte as t H 4 F j 1 P 1 u tj H 4 P 1 ~P 1 F j 1 P 1!u tj j1 t j1 t H 4 P 1 ~Q j P 2 F j 2 P 2!u tj j1 t H 4 P 1 Q j u tj j1 r a+s+ H 4 P 1 Q j u tj j1 as r by the martigale covergece theorem+ For the first term of ~21!, usig ~19! ~15!, ~10!, ad ~12! we obtai ~H 4 F 1 1 H 4! t z 1 H 4 F 1 ~t! H 4 H 4 P 1 x H 4 F 1 ~t! ~I r H F H F!P 1 x H 4 F 1 ~t! P 1 x H 4 F 1 t P 1 ~P 1 F 1 P 1!x H 4 F 1 t P 1 ~Q P 2 F 2 P 2!x H 4 F 1 t P 1 Q x H 4 F 1 t ~H 4 H 4 H F H F!P 1 Q x ~H 4 F 1 t H 4!H 4 P 1 Q x r a+s+ ~H 4 F 1 t H 4!H 4 P 1 X Q 0 because X F P 1 X Q + Thus, ~21! implies that z 1t is a R rp -valued statioary ergodic process with the followig liear process represetatio: z 1t H 4 P 1 z a+s+ z Q j u tj + (22) j1 The ergodic theorem the yields, as r, 1 z 1t z 1t H 4 P 1 1 z z P 1 H 4 r a+s+ H 4 P 1 E~z 1 z 1!P 1 H 4 H 4 P 1 Q j S uu Q j P 1 H 4 0, (23) j1

10 LIMIT THEORY FOR EXPLOSIVELY COINTEGRATED SYSTEMS 873 where positive defiiteess follows because S uu 0, P 1 has full row rak equal to r, ad H 4 has full colum rak equal to r p+ Thus, i the directio of H 4, the sample variace has the usual 1 ormalizatio that applies uder statioarity+ By stadard iversio of a partitioed matrix ~e+g+, Abadir ad Magus, 2005, 5+18! we obtai 1 z t z t z 1t z 1t z 2t z 1t z 1t z 2t z 2t z 2t 1 Z 1 Z 1 Z 1 1 Z 2 Z 2 Z 1 Z 2 Z 2 ~Z 1 Q 2 Z 1! 1 P 2 ~Z 1 Q 2 Z 1! 1 ~Z 1 Q 2 Z 1! 1 P 2 ~Z 2 Z 2! 1 P 2 ~Z 1 Q 2 Z 1! 1, (24) P 2 where Z 11, z 12,+++,z 1 # R ~rp!, Z 21 R ~Krp!, P 2 ~Z 2 Z 2! 1 Z 2 Z 1, ad Q 2 I Z 2 ~Z 2 Z 2! 1 Z 2 +, z 22 Lemma 4+4 implies that 7Z 2 Z 2 7 O p ~7Q7 2!,7P 2 7 O p ~7Q7!, ad ~ 1 Z 1 Q 2 Z 1! 1 ~ 1 Z 1 Z 1! 1 O p ~ 1!+,+++,z 2 # Thus, i view of ~23!, the large-sample behavior of the sample momet matrix after rotatio of the regressio space is give by 1 z t z t 1 1 Z 1 Q 2 Z 1 O p ~7Q7! O p ~7Q7!! O p ~7Q z 1t z 1t O p ~ 1! O p ~7Q7! O p ~7Q7! p H 4 P 1 r j1 Q j S uu Q j P 1 H 4 1 O p ~7Q7 2! (25) 0 Now that we have established a osigular limit for the sample momet matrix i the ew regressio coordiates the limit theory for the coefficiet

11 874 PETER C.B. PHILLIPS AND TASSOS MAGDALINOS matrix C i ~9! is drive by the sample covariace 102 ~z 1t «t!, which has a mixed ormal asymptotic distributio 1 ~z 1t «t! MN0,H 4 P 1 Q j S uu Q j P 1 H M j1 4 S ««(26) by virtue of a martigale cetral limit theorem+ The proof of ~26! is give i Sectio 4+ For the least squares estimator of C, combiig ~18!, ~14!, ad ~25! yields M ~ ZC C! 1 M «t z 1t 1 1 z 1t z 1t H 4, 0 o p ~1!+ It is ow straightforward to derive a limit theory for the origial coitegrated system ~1! ad ~2! by usig the relatioship AZ A ~ ZC C!P, so that M ~ AZ A! 1 Mvec~ Z M «t z 1t A A! P 1 H z 1t z 1t 1H 4 P 1 o p ~1!, 1 z 1t z 1t I m 1 ~z 1t «t! o p ~1!, M ad the limit distributio of the least squares estimator follows as a cosequece of ~23! ad ~26!+ THEOREM 2+3+ For the explosive coitegrated system geerated by (1) ad (2) with c i c j for some i j the followig limit theory applies as r : Mvec~ AZ A! MN0, P 1 H 4H 4 P 1 Remarks 2.2. j1 Q j S uu Q j P 1 H 4 1 H 4 P 1 S ««. ~i! The limit distributio of the least squares estimator is mixed Gaussia ad sigular, because rak~h 4! r p ad rak~p 1! r implies that P 1 H 4H 4 P 1 j1 Q j S uu Q j P 1 H 4 1 H 4 P 1 is a sigular matrix of rak r p+ Moreover, a M covergece rate applies, which is much slower tha the usual Q rate for explosive processes appearig i Theorem 2+2+ This reductio i the covergece rate results from the fact that some regressors i certai directios are explosively coitegrated with a commo explosive form, whereas the com-

12 [ plemetary set of regressors behave like statioary variates+ These variates slow dow the covergece rate, ad stadard limit theory applies+ ~ii! Ulike Theorem 2+2, Theorem 2+3 does ot require ay distributioal assumptios o the iovatios v t + The limitig distributio of Theorem 2+3 is valid for o-gaussia iovatios as a cosequece of the cetral limit theorem applyig for the sample covariace i ~26!+ ~iii! I the polar case where all localizig coefficiets are equal, Q ui K, r K, p 1, ad P 1 I K, ad so Theorem 2+3 reduces to LIMIT THEORY FOR EXPLOSIVELY COINTEGRATED SYSTEMS 875 u 2 1 vec~ AZ A! MN~0, H 4 ~H 4 S uu H 4! 1 H 4 S ««!+ ~iv! A iterestig feature of the limit distributio of Theorem 2+3 isthe relatioship betwee the rak of the limitig covariace matrix ad the order of coitegratio betwee the explosive regressors+ As oted i Remark 2+2~i!, the rak of the limitig covariace matrix is give by ~r p!m r i pm, p i1 where p is the umber of repeated roots of Q ad r i is the umber of times that the repeated root u i appears i Q+ Hece, the limitig covariace matrix assumes its maximum rak, ~K 1!m, whe all diagoal elemets of Q are equal+ O the other had, the iequality r 2p implies that the miimum rak, m, occurs whe r 2 ad p 1, i+e+, whe Q has exactly two equal diagoal elemets+ The rak of the limitig distributio of Theorem 2+3 reflects the fact that the orthogoal trasformatio H removes the sigularity i ~4! by cacelig out the effect of the regressors i ~2! that are ot coitegrated+ The M limit theory of Theorem 2+3 is drive exclusively from the coitegrated part of x t, i+e+, the regressors i ~2! that cotai repeated explosive roots+ ~v! I view of Theorem 2+3, the limit behavior of 1 1 x t x t z t z t HP 1 P H 1 r p P 1 H 4H 4 P 1 j1 Q j S uu Q j P 1 H 4 1 H 4 P 1, ad the fact that ZS ««1 «t «[ t r p S ««, where «[ t y t AZ x t, we obtai covetioal asymptotic chi-squared distributios uder the ull hypothesis for regressio Wald tests such as

13 Z 876 PETER C.B. PHILLIPS AND TASSOS MAGDALINOS W g~ Z A! G A 1 x t x t ZS ««1 G A g~ AZ!, G A ]g ] vec A, for some aalytic restrictios of the form H 0 : g~a! 0+ ~vi! Note that the matrix P 1 H 4H 4 P 1 j1 Q j S uu Q j P 1 H 4 1 H 4 P 1 is ivariat to the coordiate system defiig H 4, ad so the limit theory of Theorem 2+3 is also ivariat to the choice of coordiates+ We ow provide a discussio of the asymptotic behavior of AZ A i the directio of X Q + Recallig the partitioed form of P ad ~14!, the vector HPX U 4 PX Q Q Q U F PX H 4 P 1 X Q Q 0 Q U F PX ~rp!1 U F PX cacels out the effect of ~Z 1 Z 1 0! 1 o the variace matrix i ~24! ad produces a typical explosive limit theory for A + More specifically, lettig D : U F FU F, ~8!, ~18!, ad ~24! yield ~ AZ A!Q X Q ~ ZC C!F PX Q ~ ZC C!H ~HF H!HPX Q ~ ZC C!H diag~u 4 F U 4,U F F U F!HPX Q 0 «t z t z t z t 1 Q ~rp!1 D U F PX «t z 2 Z 2! 1 D P 2 ~Z 1 Q 2 Z 1! 1 P 2 D #U F PX Q «t z 1t ~Z 1 Q 2 Z 1! 1 P 2 D U F PX Q + (27) From the aalysis precedig Theorem 2+3, we kow that 7P 2 7 O p ~7Q7!, ~Z 1 Q 2 Z 1! 1 O p ~ 1!, ad «t z 1t O p ~ 102!+ Thus, because D is a diagoal matrix cosistig of all distict diagoal elemets of Q, 7D7 7Q7, ad the last term i ~27! has asymptotic order O p ~ 102!+ O the other had, usig ~16! ad the fact that U 4 F U F H 4 F 1 H F 0, we ca write

14 «t z 2t D «t x t P U F U F F U F «t x t P ~I K U 4 U 4!F U F «t x t ~P F P!P U F «t x t Q P U F, (28) so that «t z 2t O p ~7Q7!, ad the secod term i ~27! has asymptotic order O p ~ 1!+ Thus, Lemma 4+4~i! ad ~28! yield ~ Z A A!Q X Q «t z 2t D ~D Z 2 Z 2 D! 1 U F PX Q O p ~ 102! «t x t Q P U FU F PQ x t x t Q P U F 1 U F PX Q + The asymptotic behavior of AZ A i the directio of X Q is determied by a argumet idetical to the osigular case of Theorem 2+2+ For Gaussia iovatios u t, Lemma 4+2, ~33!, ad ~34! imply that Q x t x t Q ad «t x t Q coverge joitly i distributio, leadig to a mixed ormal limit, stated formally as follows+ THEOREM 2+4+ For the explosive coitegrated system geerated by (1) ad (2) with v t d N~0, S! ad c i c j for some i j the followig limit theory applies o the directio of X Q : ~ AZ A!Q X Q MN~0, X Q P U F V 1 U F PX Q S ««!, where V U F P Q j X Q X Q Q j P U F. j0 Remarks 2.3. LIMIT THEORY FOR EXPLOSIVELY COINTEGRATED SYSTEMS 877 ~i! The limit theory for the least squares estimator i the directio of X Q is mixed Gaussia with full rak covariace matrix of order m ad the usual explosive rate of covergece+ As i the osigular case, the

15 Z 878 PETER C.B. PHILLIPS AND TASSOS MAGDALINOS assumptio of Gaussia iovatios is essetial for the reasos explaied i Remark 2+1~i!+ ~ii! Rotatio of the regressio space i the directio of X Q determies the limit theory i the explosive directio resolvig the sigularity of the limitig momet matrix j0 Q j X Q X Q Q j + ~iii! I the polar case of equal localizig coefficiets, we have Q ui K with P I K + Thus, Theorem 2+4 reduces to u 1 Mu 2 1 ~ AZ A!X Q MN~0, S ««!+ 3. DISCUSSION This paper provides a limit theory for explosively coitegrated systems+ Both the ormalizatio ad the limit distributio of the cetered least squares estimate A A are foud to vary accordig to whether the regressors cotai commo explosive roots+ Whe all the explosive roots are distict, the Q expoetial rate of covergece ad a full rak mixed ormal limitig distributio apply uder the assumptio of Gaussia iovatios+ O the other had, repeated explosive roots give rise to a degeeracy i the regressio limit theory+ This degeeracy is resolved aalytically by a appropriate orthogoal rotatio of the regressio coordiates+ The resultig limit theory is mixed ormal ad of reduced rak+ The rak of the limit distributio depeds o the umber of repeated roots but is ivariat to both the choice of coordiates ad the distributio of the iovatios+ Thus, i the case where some explosive roots are commo, a ivariace priciple holds+ The authors have show that similar results to those give here hold for mildly explosive coitegrated systems with roots that approach uity at rates slower tha 1 + I particular, Magdalios ad Phillips ~2006! cosider models such as ~1! ad ~2! with mildly explosive roots of the form Q I K C a, a ~0,1!, C diag~c 1,+++,c K! 0+ For such systems, a mixed ormal asymptotic distributio is derived for the least squares estimator with the mildly explosive rate of covergece a Q whe C has distict diagoal elemets ad with the moderately statioary rate ~1a!02 whe C has repeated roots, correspodig to Theorems 2+2 ad 2+3, respectively+ A attractive feature of mildly explosive systems is that cetral limit theory applies i both cases ad asymptotic mixed ormality is valid without distributioal assumptios o the iovatios eve whe C has distict diagoal elemets+ Such systems may also be more realistic for practical work+

16 4. PROOFS LIMIT THEORY FOR EXPLOSIVELY COINTEGRATED SYSTEMS 879 This sectio cotais some techical lemmas ad also proofs of various statemets ad results i the paper+ Throughout, we use the otatio k : {02}, k X k : Q j u j, (29) j1 F t : s~v 1,+++,v t! for the atural filtratio of the iovatios, ad we let C be a boudig costat i ~0,! that may assume differet values+ The precedig choice for k is made for the sake of simplicity, ad the results hold for ay iteger-valued sequece k satisfyig 1 7Q7 2k ad 7Q7 ~k! k 102 r 0 as r + LEMMA 4+1+ For k ad X k as defied i (29), we have t max Q j u j jk a+s+ o 1 as r. 1 M Proof. Usig Doob s iequality for martigales we obtai, for each d 0, t P max Q j u j d E Q j u j 2 k 1t 1 k 1t jk 1 M 1 d 2 1 E7u 17 2 d 2 1 jk 1 C 7Q7 2k 1 7Q7 2j jk 1 C 7Q7 + 1 LEMMA 4+2+ For k ad X k as defied i (29), we have, as r, ~Q I m! ~x t «t! tk 1 ~Q t I m!~x k «t! o a+s+ ~1!. Proof. The lemma will follow by showig ~30! ad ~31!+ k ~Q t I m!@~q t x t! «t # o a+s+~1! (30)

17 880 PETER C.B. PHILLIPS AND TASSOS MAGDALINOS tk 1 ~Q t I m! t jk 1 Q j u j «t o a+s+~1!+ (31) For ~30!, because 7Q t x t 7 7Q t x t X Q 7 7X Q 7 ad 7Q t x t X Q 7 o a+s+ ~1!, there exists a costat C ~0,! such that 7Q t x t 7 C 7X Q 7 t 1 a+s+ (32) with 7X Q 7 almost surely by the martigale covergece theorem+ Thus, by ergodicity, k ~Q t I m!@~q t x t! «t # k ~C 7X Q 7!7Q7 7Q7 t 7«t 7 ~C 7X Q 7!7Q7 k 202 7Q7 O a+s+ ~7Q7 ~k! k 102! o a+s+ ~1!, k «t 7 showig ~30!+ For ~31!, Lemma 4+1 ad the ergodic theorem yield tk 1 k ~Q t I m! max k 1t t t jk 1 max k 1t t jk 1 jk 1 CM max k 1t Q j u j «t Q j u j tk 1 Q j u j t jk 1 tk 1 Q j u j 1 7Q7 t 7«t 7 7Q7 2~t!102 tk 1 Proof of Theorem 2.1. By ~32! we obtai, almost surely, 7«t 7102 tk 1 7«t 7102 o a+s+ ~1!+ k Q ~t! Q t x t x t Q t Q ~t! ~C 7X Q7! 2 7Q7 2~t! O a+s+ ~7Q7 2~k!!+

18 Thus, Lemma 4+1 yields Q x t x t Q tk 1 tk 1 Q ~t! Q t x t x t Q t Q ~t! o a+s+ ~1! Q ~t! X k X k Q ~t! o a+s+ ~1! k 1 Q j X k X k Q j o a+s+ ~1!, (33) j0 ad Theorem 2+1 follows immediately by the martigale covergece theorem+ Proof of Theorem 2.2. By ~33! ad Lemma 4+2, vec@~ AZ A!Q # 1 Q x t x t Q k 1 j1 Q j X k X k Q j0 I m~q I m! ~x t «t! I m tk 1 The assumptio of Gaussia errors yields, coditioal o F k, tk 1 ~Q t X k «t! o a+s+ ~1!+ ~Q t X k «t! d N0, j k 1 Q j X k X k Q S ««, (34) j0 which leads to vec@~ Z LIMIT THEORY FOR EXPLOSIVELY COINTEGRATED SYSTEMS 881 k 1 A A!Q # d j0 MN0, j102 Q j X k X k Q I mn~0, I K S ««! j0 j 1 Q j X Q X Q Q S ««, as r, because X k r a+s+ X Q + LEMMA 4+3+ The determiat of the matrix 1 W : ~s! w i w j 1 : i, j $1,+++,s%

19 882 PETER C.B. PHILLIPS AND TASSOS MAGDALINOS is give by 1 6W ~s! 6 ~w 2 1 1! +++~w 2 s 1! s1 ~w j w j1! 2 ~w j w j2! 2 +++~w j w s! 2 j1 ~w j w j1 1! 2 ~w j w j2 1! 2 +++~w j w s 1!. 2 Cosequetly, the matrix M W : w i w j w i w j 1 : i, j $1,+++,s% is osigular if ad oly if w i w j for all i j. Proof. We use iductio+ The result is immediate for s 2+ If we assume the result for s 1 ad partitio W ~s! as W ~s1! W ~s! w w 1 w s 2 1 w : 1 w 1 w s 1,+++, 1 we have ~e+g+, Abadir ad Magus, 2005, 5+29! ~w s 2 1!6W ~s! 6 6W ~s1! ~w s 2 1!ww 6+, w s1 w s 1 Because the matrix o the right is equal to diag w 1 w s w 1 w s 1,+++, w s1 w s w s1 w s diag 1W w ~s1! 1 w s w 1 w s 1,+++, w s1 w s ad 6W ~s1! 6 is kow from the iductio hypothesis, we obtai 6W ~s! 6 1 w s 2 1 ~w 1 w s! 2 ~w 1 w s 1! ~w s1 w s! 2 ~w s1 w s 1! 2 6W ~s1! 6 1 s1 ~w i w s! 2 ~w 2 1 1! +++~w 2 s 1! i1 ~w i w s 1! 2 s2 ~w j w j1! 2 +++~w j w s1! 2 j1 ~w j w j1 1! 2 +++~w j w s1 1! 2 1 s1 ~w j w j1! 2 +++~w j w s! 2 ~w 2 1 1! +++~w 2 s 1! j1 ~w j w j1 1! 2 +++~w j w s 1! 2 w s1 w s 1

20 LIMIT THEORY FOR EXPLOSIVELY COINTEGRATED SYSTEMS 883 as required+ Hece, W ~s! is osigular if ad oly if w i w j for all i j+ The idetity M W diag~w 1,+++,w s!w ~s! diag~w 1,+++,w s! implies that osigularity of M W is equivalet to osigularity of W ~s! + LEMMA 4+4+ Let D U F FU F. The followig results hold as r : (i) D Z 2 Z 2 D r a+s+ U F P~ j0 Q j X Q X Q Q j!p U F 0a+s+, (ii) 7Z 2 Z 1 7 O p ~7Q7!, (iii) ~Z 1 Q 2 Z 1 0! 1 ~Z 1 Z 1 0! 1 O p ~ 1!. Proof. For part ~i!, first ote that U F F j U 4 H F F 1 j H 4 0 for all j N+ Thus, usig ~17! ad ~16! we obtai D j z 2t U F F j U F U F Px t U F F j ~I K U 4 U 4!Px t U F F j Px t ~U F F j U 4!U 4 Px t U F P~P F j P!x t U F PQ j x t, (35) for all j N, ad similarly D j U F PX Q U F PQ j X Q + (36) The limit matrix of part ~i! ow follows immediately from ~35! ad Theorem 2+1: D Z 2 Z 2 D U F F U F z 2t z 2t U F F U F U F PQ x t x t Q P U F r a+s+ U F P Q j X Q X Q Q j P U F + j0 To establish the osigularity of the limit matrix, ote that, by ~5!, D U F FU F diag~u 1,+++,u p, w 1,+++,w Kr! cosists of all distict diagoal elemets of Q+ The, deotig by d i the ith diagoal elemet of D, Lemma 4+3 implies that the matrix M D : d i d j d i d j 1 : i, j $1,+++,Krp%

21 884 PETER C.B. PHILLIPS AND TASSOS MAGDALINOS ~i is osigular+ Deotig by S! F the ith elemet of the vector S F U F PX Q ad lettig SX F : diag~s ~1! F,+++,S ~Krp! F!, ~36! gives U F P Q j X Q X Q Q j P U F D j S F S F D j j0 j0 SX F M D SX F + The last matrix is osigular almost surely because M D is osigular ad S F ~i! 0 almost surely for each i by absolute cotiuity of u t + For part ~ii!, usig a matrix Cauchy Schwarz iequality ~e+g+, Abadir ad Magus, 2005, 12+5! we obtai 7z 1t 7 2 7H 4 z 7 2 tr~h 4 H 4 z 4 H 4!# 7 2 tr~z z!# 4 H 4!# 102 7z 7 2 ~r p! 102 7z 7 2 because H 4 H 4 I rp + Also, usig ~35! ad a stadard trace iequality we ca write E7z 2t 7 2 2t z 2t!# F U F!tr~ xi t xi t!# KE7xI t 7 2 KE7uI F F72t + Thus, E7vec~Z 2 Z 1!7 E~7z 1t 77z 2t 7! ~E7z 1t 7 2! 102 ~E7z 2t 7 2! 102 ~r p! 104 ~E7z 1 7 2! 102 ~E7z 2t 7 2! 102 KE7 I ~r p! 104 u 17 2 E7z F F7 t O~7F7! ad the result follows because 7F7 7Q7+ For part ~iii!, ote that part ~i! implies that 7Z 2 Z 2 7 O p ~7Q7 2!+ Thus, 1 7Z 1 Z 2 ~Z 2 Z 2! 1 Z 2 Z Z 1 Z Z 2 Z O p 1 by part ~ii!, ad the result follows from the defiitio of Q 2 +

22 LIMIT THEORY FOR EXPLOSIVELY COINTEGRATED SYSTEMS 885 Proof of (19). Because F 1 i diag~u 1 i I r1,+++,u p i I rp!, ~14! gives, for all i Z, H 4 F i 1 H F diag~u i 1 H 41 H F1,+++,u i p H 4p H Fp! 0+ O the other had, ~15! ad the precedig idetity for i 1 yield ~H 4 F 1 1 H 4! 2 H 4 F 1 1 H 4 H 4 F 1 1 H 4 H 4 F 1 1 ~I r H F H F!F 1 1 H 4 H 4 F 1 2 H 4 ~H 4 F 1 1 H F!H F F 1 1 H 4 H 4 F 1 2 H 4, ad so we have proved the idetity ~H 4 F 1 1 H 4! j H 4 F j 1 H 4 for j 2+ The geeral case follows by straightforward iductio+ Proof of (21). Usig ~10! ad ~11! the defiitio of z 1t yields z 1t H 4 P 1 x t H 4 P 1 ~Q 1 x Q 1 u! H 4 P 1 Q 1 x H 4 P 1 Q 1 u H 4 P 1 Q 1 ~P 1 P 1 P 2 P 2!x H 4 P 1 Q 1 ~P 1 P 1 P 2 P 2!u H 4 ~P 1 Q 1 P 1!P 1 x H 4 ~P 1 Q 1 P 1!P 1 u H 4 F 1 1 P 1 x H 4 F 1 1 P 1 u + The secod term has the form that appears i ~21!+ For the first term, usig the fact that H 4 F 1 H F 0, we ca write H 4 F 1 1 P 1 x H 4 F 1 1 ~H 4 H 4 H F H F!P 1 x ~H 4 F 1 1 H 4!H 4 P 1 x ~H 4 F 1 1 H 4!z 1, as required+ Proof of (26). Recallig the otatio z j1 Q j u tj, ~22! yields the followig expressio for the sample covariace: 1 ~z 1t «t! 1 ~H 4 P 1 z «t! M M ~H 4 P 1 I m! j t, (37)

23 886 PETER C.B. PHILLIPS AND TASSOS MAGDALINOS where j t : 102 ~z «t! is a martigale differece array with respect to F, because z is s~v,v t2,+++! measurable+ The coditioal variace of j t is give by E Ft ~j t j t! 1 1 Ft ~z z! «t «t z! «t «t # j Q j S uu Q 1 r a+s+ j Q j S uu Q S ««, j1 «t «t by the ergodic theorem+ Thus, provided that the Lideberg coditio E Ft ~7j t 7 2 1$7j t 7 d%! r p 0 d 0 (38) holds, Corollary 3+1 of Hall ad Heyde ~1980! ad the Cramér Wold theorem imply that j t N0, j Q j S uu Q S ««+ (39) j1 The proof of ~38! is give ext+ The proof of ~26! follows from ~37! ad ~39!+ Proof of (38). The Lideberg coditio ~38! is equivalet to 1 7«t 7 2 E Ft ~7z 7 2 1$7z 77«t 7 d 102 %! o p ~1!+ (40) Applyig the iequality 1$7z 77«t 7 d 102 % 1$7z 7 d % 1$7«t 7 d % to ~40!, we deduce that ~38! will follow if the followig terms, S 1 1 7«t 7 2 E~7z 7 2 1$7z 7 d %!, S 2 1 7«t 7 2 1$7«t 7 d %E7z 7 2,

24 LIMIT THEORY FOR EXPLOSIVELY COINTEGRATED SYSTEMS 887 are o p ~1!+ The term S 1 r 0iL 1 because, usig the fact that z is a strictly statioary sequece with E7z 1 7 2, ES 1 max E~7z 7 2 1$7z 7 d %!E7« t E~7z $7z 1 7 d %!E7«1 7 2 r 0+ The term S 2 also teds to 0 i L 1 because ES 2 E7z E~7«t 7 2 1$7«t 7 d %! E7z E~7« $7«1 7 d %! r 0, by itegrability of 7« Thus, ~40! ad ~38! follow+ REFERENCES Abadir, K+M+ &J+R+ Magus ~2005! Matrix Algebra+ Ecoometric Exercises, vol+ 1+ Cambridge Uiversity Press+ Aderso, T+W+ ~1959! O asymptotic distributios of estimates of parameters of stochastic differece equatios+ Aals of Mathematical Statistics 30, Hall, P+ &C+C+ Heyde ~1980! Martigale Limit Theory ad Its Applicatio+ Academic Press+ Magdalios, T+ &P+C+B+ Phillips ~2006! Limit Theory for Coitegrated Systems with Moderately Itegrated ad Moderately Explosive Regressors+ Workig paper, Yale Uiversity+ Mijheer, J+ ~2002! Asymptotic iferece for AR~1! processes with ~oormal! stable iovatios, part V: The explosive case+ Joural of Mathematical Scieces 111, Park, J+Y+ &P+C+B+ Phillips ~1988! Statistical iferece i regressios with itegrated processes, part 1+ Ecoometric Theory 4, Park, J+Y+ &P+C+B+ Phillips ~1989! Statistical iferece i regressios with itegrated processes, part 2+ Ecoometric Theory 5, Phillips, P+C+B+ ~1985! The distributio of matrix quotiets+ Joural of Multivariate Aalysis 16, White, J+S+ ~1958! The limitig distributio of the serial correlatio coefficiet i the explosive case+ Aals of Mathematical Statistics 29,

Regression with an Evaporating Logarithmic Trend

Regression with an Evaporating Logarithmic Trend Regressio with a Evaporatig Logarithmic Tred Peter C. B. Phillips Cowles Foudatio, Yale Uiversity, Uiversity of Aucklad & Uiversity of York ad Yixiao Su Departmet of Ecoomics Yale Uiversity October 5,

More information

Solution to Chapter 2 Analytical Exercises

Solution to Chapter 2 Analytical Exercises Nov. 25, 23, Revised Dec. 27, 23 Hayashi Ecoometrics Solutio to Chapter 2 Aalytical Exercises. For ay ε >, So, plim z =. O the other had, which meas that lim E(z =. 2. As show i the hit, Prob( z > ε =

More information

Limit Theory for Cointegrated Systems with Moderately Integrated and Moderately Explosive Regressors 1

Limit Theory for Cointegrated Systems with Moderately Integrated and Moderately Explosive Regressors 1 Limit Theory for Coitegrated Systems with Moderately Itegrated ad Moderately Explosive Regressors Tassos Magdalios Uiversity of Nottigham, UK Peter C. B. Phillips Cowles Foudatio for Research i Ecoomics

More information

Efficient GMM LECTURE 12 GMM II

Efficient GMM LECTURE 12 GMM II DECEMBER 1 010 LECTURE 1 II Efficiet The estimator depeds o the choice of the weight matrix A. The efficiet estimator is the oe that has the smallest asymptotic variace amog all estimators defied by differet

More information

Chapter 3. Strong convergence. 3.1 Definition of almost sure convergence

Chapter 3. Strong convergence. 3.1 Definition of almost sure convergence Chapter 3 Strog covergece As poited out i the Chapter 2, there are multiple ways to defie the otio of covergece of a sequece of radom variables. That chapter defied covergece i probability, covergece i

More information

10. Comparative Tests among Spatial Regression Models. Here we revisit the example in Section 8.1 of estimating the mean of a normal random

10. Comparative Tests among Spatial Regression Models. Here we revisit the example in Section 8.1 of estimating the mean of a normal random Part III. Areal Data Aalysis 0. Comparative Tests amog Spatial Regressio Models While the otio of relative likelihood values for differet models is somewhat difficult to iterpret directly (as metioed above),

More information

Apply change-of-basis formula to rewrite x as a linear combination of eigenvectors v j.

Apply change-of-basis formula to rewrite x as a linear combination of eigenvectors v j. Eigevalue-Eigevector Istructor: Nam Su Wag eigemcd Ay vector i real Euclidea space of dimesio ca be uiquely epressed as a liear combiatio of liearly idepedet vectors (ie, basis) g j, j,,, α g α g α g α

More information

Convergence of random variables. (telegram style notes) P.J.C. Spreij

Convergence of random variables. (telegram style notes) P.J.C. Spreij Covergece of radom variables (telegram style otes).j.c. Spreij this versio: September 6, 2005 Itroductio As we kow, radom variables are by defiitio measurable fuctios o some uderlyig measurable space

More information

Advanced Stochastic Processes.

Advanced Stochastic Processes. Advaced Stochastic Processes. David Gamarik LECTURE 2 Radom variables ad measurable fuctios. Strog Law of Large Numbers (SLLN). Scary stuff cotiued... Outlie of Lecture Radom variables ad measurable fuctios.

More information

Random Matrices with Blocks of Intermediate Scale Strongly Correlated Band Matrices

Random Matrices with Blocks of Intermediate Scale Strongly Correlated Band Matrices Radom Matrices with Blocks of Itermediate Scale Strogly Correlated Bad Matrices Jiayi Tog Advisor: Dr. Todd Kemp May 30, 07 Departmet of Mathematics Uiversity of Califoria, Sa Diego Cotets Itroductio Notatio

More information

The central limit theorem for Student s distribution. Problem Karim M. Abadir and Jan R. Magnus. Econometric Theory, 19, 1195 (2003)

The central limit theorem for Student s distribution. Problem Karim M. Abadir and Jan R. Magnus. Econometric Theory, 19, 1195 (2003) The cetral limit theorem for Studet s distributio Problem 03.6.1 Karim M. Abadir ad Ja R. Magus Ecoometric Theory, 19, 1195 (003) Z Ecoometric Theory, 19, 003, 1195 1198+ Prited i the Uited States of America+

More information

LECTURE 11 LINEAR PROCESSES III: ASYMPTOTIC RESULTS

LECTURE 11 LINEAR PROCESSES III: ASYMPTOTIC RESULTS PRIL 7, 9 where LECTURE LINER PROCESSES III: SYMPTOTIC RESULTS (Phillips ad Solo (99) ad Phillips Lecture Notes o Statioary ad Nostatioary Time Series) I this lecture, we discuss the LLN ad CLT for a liear

More information

Slide Set 13 Linear Model with Endogenous Regressors and the GMM estimator

Slide Set 13 Linear Model with Endogenous Regressors and the GMM estimator Slide Set 13 Liear Model with Edogeous Regressors ad the GMM estimator Pietro Coretto pcoretto@uisa.it Ecoometrics Master i Ecoomics ad Fiace (MEF) Uiversità degli Studi di Napoli Federico II Versio: Friday

More information

CEE 522 Autumn Uncertainty Concepts for Geotechnical Engineering

CEE 522 Autumn Uncertainty Concepts for Geotechnical Engineering CEE 5 Autum 005 Ucertaity Cocepts for Geotechical Egieerig Basic Termiology Set A set is a collectio of (mutually exclusive) objects or evets. The sample space is the (collectively exhaustive) collectio

More information

CALCULATION OF FIBONACCI VECTORS

CALCULATION OF FIBONACCI VECTORS CALCULATION OF FIBONACCI VECTORS Stuart D. Aderso Departmet of Physics, Ithaca College 953 Daby Road, Ithaca NY 14850, USA email: saderso@ithaca.edu ad Dai Novak Departmet of Mathematics, Ithaca College

More information

Berry-Esseen bounds for self-normalized martingales

Berry-Esseen bounds for self-normalized martingales Berry-Essee bouds for self-ormalized martigales Xiequa Fa a, Qi-Ma Shao b a Ceter for Applied Mathematics, Tiaji Uiversity, Tiaji 30007, Chia b Departmet of Statistics, The Chiese Uiversity of Hog Kog,

More information

Statistical Inference Based on Extremum Estimators

Statistical Inference Based on Extremum Estimators T. Rotheberg Fall, 2007 Statistical Iferece Based o Extremum Estimators Itroductio Suppose 0, the true value of a p-dimesioal parameter, is kow to lie i some subset S R p : Ofte we choose to estimate 0

More information

4. Partial Sums and the Central Limit Theorem

4. Partial Sums and the Central Limit Theorem 1 of 10 7/16/2009 6:05 AM Virtual Laboratories > 6. Radom Samples > 1 2 3 4 5 6 7 4. Partial Sums ad the Cetral Limit Theorem The cetral limit theorem ad the law of large umbers are the two fudametal theorems

More information

UNIT ROOT MODEL SELECTION PETER C. B. PHILLIPS COWLES FOUNDATION PAPER NO. 1231

UNIT ROOT MODEL SELECTION PETER C. B. PHILLIPS COWLES FOUNDATION PAPER NO. 1231 UNIT ROOT MODEL SELECTION BY PETER C. B. PHILLIPS COWLES FOUNDATION PAPER NO. 1231 COWLES FOUNDATION FOR RESEARCH IN ECONOMICS YALE UNIVERSITY Box 28281 New Have, Coecticut 652-8281 28 http://cowles.eco.yale.edu/

More information

17. Joint distributions of extreme order statistics Lehmann 5.1; Ferguson 15

17. Joint distributions of extreme order statistics Lehmann 5.1; Ferguson 15 17. Joit distributios of extreme order statistics Lehma 5.1; Ferguso 15 I Example 10., we derived the asymptotic distributio of the maximum from a radom sample from a uiform distributio. We did this usig

More information

Diagonal approximations by martingales

Diagonal approximations by martingales Alea 7, 257 276 200 Diagoal approximatios by martigales Jaa Klicarová ad Dalibor Volý Faculty of Ecoomics, Uiversity of South Bohemia, Studetsa 3, 370 05, Cese Budejovice, Czech Republic E-mail address:

More information

Notes 19 : Martingale CLT

Notes 19 : Martingale CLT Notes 9 : Martigale CLT Math 733-734: Theory of Probability Lecturer: Sebastie Roch Refereces: [Bil95, Chapter 35], [Roc, Chapter 3]. Sice we have ot ecoutered weak covergece i some time, we first recall

More information

A Weak Law of Large Numbers Under Weak Mixing

A Weak Law of Large Numbers Under Weak Mixing A Weak Law of Large Numbers Uder Weak Mixig Bruce E. Hase Uiversity of Wiscosi Jauary 209 Abstract This paper presets a ew weak law of large umbers (WLLN) for heterogeous depedet processes ad arrays. The

More information

Lecture 7: Properties of Random Samples

Lecture 7: Properties of Random Samples Lecture 7: Properties of Radom Samples 1 Cotiued From Last Class Theorem 1.1. Let X 1, X,...X be a radom sample from a populatio with mea µ ad variace σ

More information

MA Advanced Econometrics: Properties of Least Squares Estimators

MA Advanced Econometrics: Properties of Least Squares Estimators MA Advaced Ecoometrics: Properties of Least Squares Estimators Karl Whela School of Ecoomics, UCD February 5, 20 Karl Whela UCD Least Squares Estimators February 5, 20 / 5 Part I Least Squares: Some Fiite-Sample

More information

Kolmogorov-Smirnov type Tests for Local Gaussianity in High-Frequency Data

Kolmogorov-Smirnov type Tests for Local Gaussianity in High-Frequency Data Proceedigs 59th ISI World Statistics Cogress, 5-30 August 013, Hog Kog (Sessio STS046) p.09 Kolmogorov-Smirov type Tests for Local Gaussiaity i High-Frequecy Data George Tauche, Duke Uiversity Viktor Todorov,

More information

MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.436J/15.085J Fall 2008 Lecture 19 11/17/2008 LAWS OF LARGE NUMBERS II THE STRONG LAW OF LARGE NUMBERS

MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.436J/15.085J Fall 2008 Lecture 19 11/17/2008 LAWS OF LARGE NUMBERS II THE STRONG LAW OF LARGE NUMBERS MASSACHUSTTS INSTITUT OF TCHNOLOGY 6.436J/5.085J Fall 2008 Lecture 9 /7/2008 LAWS OF LARG NUMBRS II Cotets. The strog law of large umbers 2. The Cheroff boud TH STRONG LAW OF LARG NUMBRS While the weak

More information

Lecture 19: Convergence

Lecture 19: Convergence Lecture 19: Covergece Asymptotic approach I statistical aalysis or iferece, a key to the success of fidig a good procedure is beig able to fid some momets ad/or distributios of various statistics. I may

More information

Hoggatt and King [lo] defined a complete sequence of natural numbers

Hoggatt and King [lo] defined a complete sequence of natural numbers REPRESENTATIONS OF N AS A SUM OF DISTINCT ELEMENTS FROM SPECIAL SEQUENCES DAVID A. KLARNER, Uiversity of Alberta, Edmoto, Caada 1. INTRODUCTION Let a, I deote a sequece of atural umbers which satisfies

More information

LECTURE 8: ASYMPTOTICS I

LECTURE 8: ASYMPTOTICS I LECTURE 8: ASYMPTOTICS I We are iterested i the properties of estimators as. Cosider a sequece of radom variables {, X 1}. N. M. Kiefer, Corell Uiversity, Ecoomics 60 1 Defiitio: (Weak covergece) A sequece

More information

LECTURE 8: ORTHOGONALITY (CHAPTER 5 IN THE BOOK)

LECTURE 8: ORTHOGONALITY (CHAPTER 5 IN THE BOOK) LECTURE 8: ORTHOGONALITY (CHAPTER 5 IN THE BOOK) Everythig marked by is ot required by the course syllabus I this lecture, all vector spaces is over the real umber R. All vectors i R is viewed as a colum

More information

Definition 4.2. (a) A sequence {x n } in a Banach space X is a basis for X if. unique scalars a n (x) such that x = n. a n (x) x n. (4.

Definition 4.2. (a) A sequence {x n } in a Banach space X is a basis for X if. unique scalars a n (x) such that x = n. a n (x) x n. (4. 4. BASES I BAACH SPACES 39 4. BASES I BAACH SPACES Sice a Baach space X is a vector space, it must possess a Hamel, or vector space, basis, i.e., a subset {x γ } γ Γ whose fiite liear spa is all of X ad

More information

ANOTHER GENERALIZED FIBONACCI SEQUENCE 1. INTRODUCTION

ANOTHER GENERALIZED FIBONACCI SEQUENCE 1. INTRODUCTION ANOTHER GENERALIZED FIBONACCI SEQUENCE MARCELLUS E. WADDILL A N D LOUIS SACKS Wake Forest College, Wisto Salem, N. C., ad Uiversity of ittsburgh, ittsburgh, a. 1. INTRODUCTION Recet issues of umerous periodicals

More information

Chimica Inorganica 3

Chimica Inorganica 3 himica Iorgaica Irreducible Represetatios ad haracter Tables Rather tha usig geometrical operatios, it is ofte much more coveiet to employ a ew set of group elemets which are matrices ad to make the rule

More information

A Hadamard-type lower bound for symmetric diagonally dominant positive matrices

A Hadamard-type lower bound for symmetric diagonally dominant positive matrices A Hadamard-type lower boud for symmetric diagoally domiat positive matrices Christopher J. Hillar, Adre Wibisoo Uiversity of Califoria, Berkeley Jauary 7, 205 Abstract We prove a ew lower-boud form of

More information

KU Leuven Department of Computer Science

KU Leuven Department of Computer Science O orthogoal polyomials related to arithmetic ad harmoic sequeces Adhemar Bultheel ad Adreas Lasarow Report TW 687, February 208 KU Leuve Departmet of Computer Sciece Celestijelaa 200A B-300 Heverlee (Belgium)

More information

5 Birkhoff s Ergodic Theorem

5 Birkhoff s Ergodic Theorem 5 Birkhoff s Ergodic Theorem Amog the most useful of the various geeralizatios of KolmogorovâĂŹs strog law of large umbers are the ergodic theorems of Birkhoff ad Kigma, which exted the validity of the

More information

Infinite Sequences and Series

Infinite Sequences and Series Chapter 6 Ifiite Sequeces ad Series 6.1 Ifiite Sequeces 6.1.1 Elemetary Cocepts Simply speakig, a sequece is a ordered list of umbers writte: {a 1, a 2, a 3,...a, a +1,...} where the elemets a i represet

More information

R is a scalar defined as follows:

R is a scalar defined as follows: Math 8. Notes o Dot Product, Cross Product, Plaes, Area, ad Volumes This lecture focuses primarily o the dot product ad its may applicatios, especially i the measuremet of agles ad scalar projectio ad

More information

A Hilbert Space Central Limit Theorem for Geometrically Ergodic Markov Chains

A Hilbert Space Central Limit Theorem for Geometrically Ergodic Markov Chains A Hilbert Space Cetral Limit Theorem for Geometrically Ergodic Marov Chais Joh Stachursi Research School of Ecoomics, Australia Natioal Uiversity Abstract This ote proves a simple but useful cetral limit

More information

Random assignment with integer costs

Random assignment with integer costs Radom assigmet with iteger costs Robert Parviaie Departmet of Mathematics, Uppsala Uiversity P.O. Box 480, SE-7506 Uppsala, Swede robert.parviaie@math.uu.se Jue 4, 200 Abstract The radom assigmet problem

More information

62. Power series Definition 16. (Power series) Given a sequence {c n }, the series. c n x n = c 0 + c 1 x + c 2 x 2 + c 3 x 3 +

62. Power series Definition 16. (Power series) Given a sequence {c n }, the series. c n x n = c 0 + c 1 x + c 2 x 2 + c 3 x 3 + 62. Power series Defiitio 16. (Power series) Give a sequece {c }, the series c x = c 0 + c 1 x + c 2 x 2 + c 3 x 3 + is called a power series i the variable x. The umbers c are called the coefficiets of

More information

Economics 241B Relation to Method of Moments and Maximum Likelihood OLSE as a Maximum Likelihood Estimator

Economics 241B Relation to Method of Moments and Maximum Likelihood OLSE as a Maximum Likelihood Estimator Ecoomics 24B Relatio to Method of Momets ad Maximum Likelihood OLSE as a Maximum Likelihood Estimator Uder Assumptio 5 we have speci ed the distributio of the error, so we ca estimate the model parameters

More information

arxiv: v1 [math.pr] 4 Dec 2013

arxiv: v1 [math.pr] 4 Dec 2013 Squared-Norm Empirical Process i Baach Space arxiv:32005v [mathpr] 4 Dec 203 Vicet Q Vu Departmet of Statistics The Ohio State Uiversity Columbus, OH vqv@statosuedu Abstract Jig Lei Departmet of Statistics

More information

NANYANG TECHNOLOGICAL UNIVERSITY SYLLABUS FOR ENTRANCE EXAMINATION FOR INTERNATIONAL STUDENTS AO-LEVEL MATHEMATICS

NANYANG TECHNOLOGICAL UNIVERSITY SYLLABUS FOR ENTRANCE EXAMINATION FOR INTERNATIONAL STUDENTS AO-LEVEL MATHEMATICS NANYANG TECHNOLOGICAL UNIVERSITY SYLLABUS FOR ENTRANCE EXAMINATION FOR INTERNATIONAL STUDENTS AO-LEVEL MATHEMATICS STRUCTURE OF EXAMINATION PAPER. There will be oe 2-hour paper cosistig of 4 questios.

More information

Geometry of LS. LECTURE 3 GEOMETRY OF LS, PROPERTIES OF σ 2, PARTITIONED REGRESSION, GOODNESS OF FIT

Geometry of LS. LECTURE 3 GEOMETRY OF LS, PROPERTIES OF σ 2, PARTITIONED REGRESSION, GOODNESS OF FIT OCTOBER 7, 2016 LECTURE 3 GEOMETRY OF LS, PROPERTIES OF σ 2, PARTITIONED REGRESSION, GOODNESS OF FIT Geometry of LS We ca thik of y ad the colums of X as members of the -dimesioal Euclidea space R Oe ca

More information

Math 61CM - Solutions to homework 3

Math 61CM - Solutions to homework 3 Math 6CM - Solutios to homework 3 Cédric De Groote October 2 th, 208 Problem : Let F be a field, m 0 a fixed oegative iteger ad let V = {a 0 + a x + + a m x m a 0,, a m F} be the vector space cosistig

More information

7.1 Convergence of sequences of random variables

7.1 Convergence of sequences of random variables Chapter 7 Limit Theorems Throughout this sectio we will assume a probability space (, F, P), i which is defied a ifiite sequece of radom variables (X ) ad a radom variable X. The fact that for every ifiite

More information

TEACHER CERTIFICATION STUDY GUIDE

TEACHER CERTIFICATION STUDY GUIDE COMPETENCY 1. ALGEBRA SKILL 1.1 1.1a. ALGEBRAIC STRUCTURES Kow why the real ad complex umbers are each a field, ad that particular rigs are ot fields (e.g., itegers, polyomial rigs, matrix rigs) Algebra

More information

1 1 2 = show that: over variables x and y. [2 marks] Write down necessary conditions involving first and second-order partial derivatives for ( x0, y

1 1 2 = show that: over variables x and y. [2 marks] Write down necessary conditions involving first and second-order partial derivatives for ( x0, y Questio (a) A square matrix A= A is called positive defiite if the quadratic form waw > 0 for every o-zero vector w [Note: Here (.) deotes the traspose of a matrix or a vector]. Let 0 A = 0 = show that:

More information

EE 4TM4: Digital Communications II Probability Theory

EE 4TM4: Digital Communications II Probability Theory 1 EE 4TM4: Digital Commuicatios II Probability Theory I. RANDOM VARIABLES A radom variable is a real-valued fuctio defied o the sample space. Example: Suppose that our experimet cosists of tossig two fair

More information

Linear regression. Daniel Hsu (COMS 4771) (y i x T i β)2 2πσ. 2 2σ 2. 1 n. (x T i β y i ) 2. 1 ˆβ arg min. β R n d

Linear regression. Daniel Hsu (COMS 4771) (y i x T i β)2 2πσ. 2 2σ 2. 1 n. (x T i β y i ) 2. 1 ˆβ arg min. β R n d Liear regressio Daiel Hsu (COMS 477) Maximum likelihood estimatio Oe of the simplest liear regressio models is the followig: (X, Y ),..., (X, Y ), (X, Y ) are iid radom pairs takig values i R d R, ad Y

More information

A GENERALIZATION OF THE SYMMETRY BETWEEN COMPLETE AND ELEMENTARY SYMMETRIC FUNCTIONS. Mircea Merca

A GENERALIZATION OF THE SYMMETRY BETWEEN COMPLETE AND ELEMENTARY SYMMETRIC FUNCTIONS. Mircea Merca Idia J Pure Appl Math 45): 75-89 February 204 c Idia Natioal Sciece Academy A GENERALIZATION OF THE SYMMETRY BETWEEN COMPLETE AND ELEMENTARY SYMMETRIC FUNCTIONS Mircea Merca Departmet of Mathematics Uiversity

More information

Product measures, Tonelli s and Fubini s theorems For use in MAT3400/4400, autumn 2014 Nadia S. Larsen. Version of 13 October 2014.

Product measures, Tonelli s and Fubini s theorems For use in MAT3400/4400, autumn 2014 Nadia S. Larsen. Version of 13 October 2014. Product measures, Toelli s ad Fubii s theorems For use i MAT3400/4400, autum 2014 Nadia S. Larse Versio of 13 October 2014. 1. Costructio of the product measure The purpose of these otes is to preset the

More information

This section is optional.

This section is optional. 4 Momet Geeratig Fuctios* This sectio is optioal. The momet geeratig fuctio g : R R of a radom variable X is defied as g(t) = E[e tx ]. Propositio 1. We have g () (0) = E[X ] for = 1, 2,... Proof. Therefore

More information

Limit Theorems. Convergence in Probability. Let X be the number of heads observed in n tosses. Then, E[X] = np and Var[X] = np(1-p).

Limit Theorems. Convergence in Probability. Let X be the number of heads observed in n tosses. Then, E[X] = np and Var[X] = np(1-p). Limit Theorems Covergece i Probability Let X be the umber of heads observed i tosses. The, E[X] = p ad Var[X] = p(-p). L O This P x p NM QP P x p should be close to uity for large if our ituitio is correct.

More information

Complex Analysis Spring 2001 Homework I Solution

Complex Analysis Spring 2001 Homework I Solution Complex Aalysis Sprig 2001 Homework I Solutio 1. Coway, Chapter 1, sectio 3, problem 3. Describe the set of poits satisfyig the equatio z a z + a = 2c, where c > 0 ad a R. To begi, we see from the triagle

More information

Chapter 6 Infinite Series

Chapter 6 Infinite Series Chapter 6 Ifiite Series I the previous chapter we cosidered itegrals which were improper i the sese that the iterval of itegratio was ubouded. I this chapter we are goig to discuss a topic which is somewhat

More information

arxiv: v1 [math.pr] 13 Oct 2011

arxiv: v1 [math.pr] 13 Oct 2011 A tail iequality for quadratic forms of subgaussia radom vectors Daiel Hsu, Sham M. Kakade,, ad Tog Zhag 3 arxiv:0.84v math.pr] 3 Oct 0 Microsoft Research New Eglad Departmet of Statistics, Wharto School,

More information

It is always the case that unions, intersections, complements, and set differences are preserved by the inverse image of a function.

It is always the case that unions, intersections, complements, and set differences are preserved by the inverse image of a function. MATH 532 Measurable Fuctios Dr. Neal, WKU Throughout, let ( X, F, µ) be a measure space ad let (!, F, P ) deote the special case of a probability space. We shall ow begi to study real-valued fuctios defied

More information

Notes 27 : Brownian motion: path properties

Notes 27 : Brownian motion: path properties Notes 27 : Browia motio: path properties Math 733-734: Theory of Probability Lecturer: Sebastie Roch Refereces:[Dur10, Sectio 8.1], [MP10, Sectio 1.1, 1.2, 1.3]. Recall: DEF 27.1 (Covariace) Let X = (X

More information

Bertrand s Postulate

Bertrand s Postulate Bertrad s Postulate Lola Thompso Ross Program July 3, 2009 Lola Thompso (Ross Program Bertrad s Postulate July 3, 2009 1 / 33 Bertrad s Postulate I ve said it oce ad I ll say it agai: There s always a

More information

Exercise 4.3 Use the Continuity Theorem to prove the Cramér-Wold Theorem, Theorem. (1) φ a X(1).

Exercise 4.3 Use the Continuity Theorem to prove the Cramér-Wold Theorem, Theorem. (1) φ a X(1). Assigmet 7 Exercise 4.3 Use the Cotiuity Theorem to prove the Cramér-Wold Theorem, Theorem 4.12. Hit: a X d a X implies that φ a X (1) φ a X(1). Sketch of solutio: As we poited out i class, the oly tricky

More information

A RANK STATISTIC FOR NON-PARAMETRIC K-SAMPLE AND CHANGE POINT PROBLEMS

A RANK STATISTIC FOR NON-PARAMETRIC K-SAMPLE AND CHANGE POINT PROBLEMS J. Japa Statist. Soc. Vol. 41 No. 1 2011 67 73 A RANK STATISTIC FOR NON-PARAMETRIC K-SAMPLE AND CHANGE POINT PROBLEMS Yoichi Nishiyama* We cosider k-sample ad chage poit problems for idepedet data i a

More information

Summary and Discussion on Simultaneous Analysis of Lasso and Dantzig Selector

Summary and Discussion on Simultaneous Analysis of Lasso and Dantzig Selector Summary ad Discussio o Simultaeous Aalysis of Lasso ad Datzig Selector STAT732, Sprig 28 Duzhe Wag May 4, 28 Abstract This is a discussio o the work i Bickel, Ritov ad Tsybakov (29). We begi with a short

More information

Session 5. (1) Principal component analysis and Karhunen-Loève transformation

Session 5. (1) Principal component analysis and Karhunen-Loève transformation 200 Autum semester Patter Iformatio Processig Topic 2 Image compressio by orthogoal trasformatio Sessio 5 () Pricipal compoet aalysis ad Karhue-Loève trasformatio Topic 2 of this course explais the image

More information

Journal of Multivariate Analysis. Superefficient estimation of the marginals by exploiting knowledge on the copula

Journal of Multivariate Analysis. Superefficient estimation of the marginals by exploiting knowledge on the copula Joural of Multivariate Aalysis 102 (2011) 1315 1319 Cotets lists available at ScieceDirect Joural of Multivariate Aalysis joural homepage: www.elsevier.com/locate/jmva Superefficiet estimatio of the margials

More information

Determinants of order 2 and 3 were defined in Chapter 2 by the formulae (5.1)

Determinants of order 2 and 3 were defined in Chapter 2 by the formulae (5.1) 5. Determiats 5.. Itroductio 5.2. Motivatio for the Choice of Axioms for a Determiat Fuctios 5.3. A Set of Axioms for a Determiat Fuctio 5.4. The Determiat of a Diagoal Matrix 5.5. The Determiat of a Upper

More information

Singular Continuous Measures by Michael Pejic 5/14/10

Singular Continuous Measures by Michael Pejic 5/14/10 Sigular Cotiuous Measures by Michael Peic 5/4/0 Prelimiaries Give a set X, a σ-algebra o X is a collectio of subsets of X that cotais X ad ad is closed uder complemetatio ad coutable uios hece, coutable

More information

Detailed proofs of Propositions 3.1 and 3.2

Detailed proofs of Propositions 3.1 and 3.2 Detailed proofs of Propositios 3. ad 3. Proof of Propositio 3. NB: itegratio sets are geerally omitted for itegrals defied over a uit hypercube [0, s with ay s d. We first give four lemmas. The proof of

More information

Chandrasekhar Type Algorithms. for the Riccati Equation of Lainiotis Filter

Chandrasekhar Type Algorithms. for the Riccati Equation of Lainiotis Filter Cotemporary Egieerig Scieces, Vol. 3, 00, o. 4, 9-00 Chadrasekhar ype Algorithms for the Riccati Equatio of Laiiotis Filter Nicholas Assimakis Departmet of Electroics echological Educatioal Istitute of

More information

Asymptotic distribution of products of sums of independent random variables

Asymptotic distribution of products of sums of independent random variables Proc. Idia Acad. Sci. Math. Sci. Vol. 3, No., May 03, pp. 83 9. c Idia Academy of Scieces Asymptotic distributio of products of sums of idepedet radom variables YANLING WANG, SUXIA YAO ad HONGXIA DU ollege

More information

1 General linear Model Continued..

1 General linear Model Continued.. Geeral liear Model Cotiued.. We have We kow y = X + u X o radom u v N(0; I ) b = (X 0 X) X 0 y E( b ) = V ar( b ) = (X 0 X) We saw that b = (X 0 X) X 0 u so b is a liear fuctio of a ormally distributed

More information

First Year Quantitative Comp Exam Spring, Part I - 203A. f X (x) = 0 otherwise

First Year Quantitative Comp Exam Spring, Part I - 203A. f X (x) = 0 otherwise First Year Quatitative Comp Exam Sprig, 2012 Istructio: There are three parts. Aswer every questio i every part. Questio I-1 Part I - 203A A radom variable X is distributed with the margial desity: >

More information

6 Integers Modulo n. integer k can be written as k = qn + r, with q,r, 0 r b. So any integer.

6 Integers Modulo n. integer k can be written as k = qn + r, with q,r, 0 r b. So any integer. 6 Itegers Modulo I Example 2.3(e), we have defied the cogruece of two itegers a,b with respect to a modulus. Let us recall that a b (mod ) meas a b. We have proved that cogruece is a equivalece relatio

More information

Math 2784 (or 2794W) University of Connecticut

Math 2784 (or 2794W) University of Connecticut ORDERS OF GROWTH PAT SMITH Math 2784 (or 2794W) Uiversity of Coecticut Date: Mar. 2, 22. ORDERS OF GROWTH. Itroductio Gaiig a ituitive feel for the relative growth of fuctios is importat if you really

More information

ECE 330:541, Stochastic Signals and Systems Lecture Notes on Limit Theorems from Probability Fall 2002

ECE 330:541, Stochastic Signals and Systems Lecture Notes on Limit Theorems from Probability Fall 2002 ECE 330:541, Stochastic Sigals ad Systems Lecture Notes o Limit Theorems from robability Fall 00 I practice, there are two ways we ca costruct a ew sequece of radom variables from a old sequece of radom

More information

Section 4.3. Boolean functions

Section 4.3. Boolean functions Sectio 4.3. Boolea fuctios Let us take aother look at the simplest o-trivial Boolea algebra, ({0}), the power-set algebra based o a oe-elemet set, chose here as {0}. This has two elemets, the empty set,

More information

The inverse eigenvalue problem for symmetric doubly stochastic matrices

The inverse eigenvalue problem for symmetric doubly stochastic matrices Liear Algebra ad its Applicatios 379 (004) 77 83 www.elsevier.com/locate/laa The iverse eigevalue problem for symmetric doubly stochastic matrices Suk-Geu Hwag a,,, Sug-Soo Pyo b, a Departmet of Mathematics

More information

PAPER : IIT-JAM 2010

PAPER : IIT-JAM 2010 MATHEMATICS-MA (CODE A) Q.-Q.5: Oly oe optio is correct for each questio. Each questio carries (+6) marks for correct aswer ad ( ) marks for icorrect aswer.. Which of the followig coditios does NOT esure

More information

Seunghee Ye Ma 8: Week 5 Oct 28

Seunghee Ye Ma 8: Week 5 Oct 28 Week 5 Summary I Sectio, we go over the Mea Value Theorem ad its applicatios. I Sectio 2, we will recap what we have covered so far this term. Topics Page Mea Value Theorem. Applicatios of the Mea Value

More information

Topic 9: Sampling Distributions of Estimators

Topic 9: Sampling Distributions of Estimators Topic 9: Samplig Distributios of Estimators Course 003, 2018 Page 0 Samplig distributios of estimators Sice our estimators are statistics (particular fuctios of radom variables), their distributio ca be

More information

Cov(aX, cy ) Var(X) Var(Y ) It is completely invariant to affine transformations: for any a, b, c, d R, ρ(ax + b, cy + d) = a.s. X i. as n.

Cov(aX, cy ) Var(X) Var(Y ) It is completely invariant to affine transformations: for any a, b, c, d R, ρ(ax + b, cy + d) = a.s. X i. as n. CS 189 Itroductio to Machie Learig Sprig 218 Note 11 1 Caoical Correlatio Aalysis The Pearso Correlatio Coefficiet ρ(x, Y ) is a way to measure how liearly related (i other words, how well a liear model

More information

2.2. Central limit theorem.

2.2. Central limit theorem. 36.. Cetral limit theorem. The most ideal case of the CLT is that the radom variables are iid with fiite variace. Although it is a special case of the more geeral Lideberg-Feller CLT, it is most stadard

More information

Stochastic Matrices in a Finite Field

Stochastic Matrices in a Finite Field Stochastic Matrices i a Fiite Field Abstract: I this project we will explore the properties of stochastic matrices i both the real ad the fiite fields. We first explore what properties 2 2 stochastic matrices

More information

COWLES FOUNDATION FOR RESEARCH IN ECONOMICS YALE UNIVERSITY

COWLES FOUNDATION FOR RESEARCH IN ECONOMICS YALE UNIVERSITY LIMIT THEORY FOR MODERATE DEVIATIONS FROM A UNIT ROOT By Peter C.B. Phillips ad Tassos Magdalios July 4 COWLES FOUNDATION DISCUSSION PAPER NO. 47 COWLES FOUNDATION FOR RESEARCH IN ECONOMICS YALE UNIVERSITY

More information

MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.265/15.070J Fall 2013 Lecture 21 11/27/2013

MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.265/15.070J Fall 2013 Lecture 21 11/27/2013 MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.265/15.070J Fall 2013 Lecture 21 11/27/2013 Fuctioal Law of Large Numbers. Costructio of the Wieer Measure Cotet. 1. Additioal techical results o weak covergece

More information

ECON 3150/4150, Spring term Lecture 3

ECON 3150/4150, Spring term Lecture 3 Itroductio Fidig the best fit by regressio Residuals ad R-sq Regressio ad causality Summary ad ext step ECON 3150/4150, Sprig term 2014. Lecture 3 Ragar Nymoe Uiversity of Oslo 21 Jauary 2014 1 / 30 Itroductio

More information

Sequences and Series of Functions

Sequences and Series of Functions Chapter 6 Sequeces ad Series of Fuctios 6.1. Covergece of a Sequece of Fuctios Poitwise Covergece. Defiitio 6.1. Let, for each N, fuctio f : A R be defied. If, for each x A, the sequece (f (x)) coverges

More information

Asymptotic Results for the Linear Regression Model

Asymptotic Results for the Linear Regression Model Asymptotic Results for the Liear Regressio Model C. Fli November 29, 2000 1. Asymptotic Results uder Classical Assumptios The followig results apply to the liear regressio model y = Xβ + ε, where X is

More information

Chapter 6 Principles of Data Reduction

Chapter 6 Principles of Data Reduction Chapter 6 for BST 695: Special Topics i Statistical Theory. Kui Zhag, 0 Chapter 6 Priciples of Data Reductio Sectio 6. Itroductio Goal: To summarize or reduce the data X, X,, X to get iformatio about a

More information

Random Walks on Discrete and Continuous Circles. by Jeffrey S. Rosenthal School of Mathematics, University of Minnesota, Minneapolis, MN, U.S.A.

Random Walks on Discrete and Continuous Circles. by Jeffrey S. Rosenthal School of Mathematics, University of Minnesota, Minneapolis, MN, U.S.A. Radom Walks o Discrete ad Cotiuous Circles by Jeffrey S. Rosethal School of Mathematics, Uiversity of Miesota, Mieapolis, MN, U.S.A. 55455 (Appeared i Joural of Applied Probability 30 (1993), 780 789.)

More information

Theorem: Let A n n. In this case that A does reduce to I, we search for A 1 as the solution matrix X to the matrix equation A X = I i.e.

Theorem: Let A n n. In this case that A does reduce to I, we search for A 1 as the solution matrix X to the matrix equation A X = I i.e. Theorem: Let A be a square matrix The A has a iverse matrix if ad oly if its reduced row echelo form is the idetity I this case the algorithm illustrated o the previous page will always yield the iverse

More information

A note on self-normalized Dickey-Fuller test for unit root in autoregressive time series with GARCH errors

A note on self-normalized Dickey-Fuller test for unit root in autoregressive time series with GARCH errors Appl. Math. J. Chiese Uiv. 008, 3(): 97-0 A ote o self-ormalized Dickey-Fuller test for uit root i autoregressive time series with GARCH errors YANG Xiao-rog ZHANG Li-xi Abstract. I this article, the uit

More information

THE SPECTRAL RADII AND NORMS OF LARGE DIMENSIONAL NON-CENTRAL RANDOM MATRICES

THE SPECTRAL RADII AND NORMS OF LARGE DIMENSIONAL NON-CENTRAL RANDOM MATRICES COMMUN. STATIST.-STOCHASTIC MODELS, 0(3), 525-532 (994) THE SPECTRAL RADII AND NORMS OF LARGE DIMENSIONAL NON-CENTRAL RANDOM MATRICES Jack W. Silverstei Departmet of Mathematics, Box 8205 North Carolia

More information

Empirical Processes: Glivenko Cantelli Theorems

Empirical Processes: Glivenko Cantelli Theorems Empirical Processes: Gliveko Catelli Theorems Mouliath Baerjee Jue 6, 200 Gliveko Catelli classes of fuctios The reader is referred to Chapter.6 of Weller s Torgo otes, Chapter??? of VDVW ad Chapter 8.3

More information

REGULARIZATION OF CERTAIN DIVERGENT SERIES OF POLYNOMIALS

REGULARIZATION OF CERTAIN DIVERGENT SERIES OF POLYNOMIALS REGULARIZATION OF CERTAIN DIVERGENT SERIES OF POLYNOMIALS LIVIU I. NICOLAESCU ABSTRACT. We ivestigate the geeralized covergece ad sums of series of the form P at P (x, where P R[x], a R,, ad T : R[x] R[x]

More information

ECONOMETRIC THEORY. MODULE XIII Lecture - 34 Asymptotic Theory and Stochastic Regressors

ECONOMETRIC THEORY. MODULE XIII Lecture - 34 Asymptotic Theory and Stochastic Regressors ECONOMETRIC THEORY MODULE XIII Lecture - 34 Asymptotic Theory ad Stochastic Regressors Dr. Shalabh Departmet of Mathematics ad Statistics Idia Istitute of Techology Kapur Asymptotic theory The asymptotic

More information

AH Checklist (Unit 3) AH Checklist (Unit 3) Matrices

AH Checklist (Unit 3) AH Checklist (Unit 3) Matrices AH Checklist (Uit 3) AH Checklist (Uit 3) Matrices Skill Achieved? Kow that a matrix is a rectagular array of umbers (aka etries or elemets) i paretheses, each etry beig i a particular row ad colum Kow

More information

SOME TRIBONACCI IDENTITIES

SOME TRIBONACCI IDENTITIES Mathematics Today Vol.7(Dec-011) 1-9 ISSN 0976-38 Abstract: SOME TRIBONACCI IDENTITIES Shah Devbhadra V. Sir P.T.Sarvajaik College of Sciece, Athwalies, Surat 395001. e-mail : drdvshah@yahoo.com The sequece

More information