Appendices to the paper "Detecting Big Structural Breaks in Large Factor Models" (2013) by Chen, Dolado and Gonzalo.
|
|
- Bathsheba Bradley
- 5 years ago
- Views:
Transcription
1 Appendices to the paper "Detecting Big Structural Breaks in Large Factor Models" 203 by Chen, Dolado and Gonzalo. A.: Proof of Propositions and 2 he proof proceeds by showing that the errors, factors and loadings in model 5 satisfy Assumptions A to D of Bai and g 2002 B 2002 hereafter. hen, once these results are proven, Propositions and 2 just follow immediately from application of heorems and 2 of B Define F t = [F t G t ], ɛ t = HG 2 t + e t, and Γ = [A Lemma. E Ft 4 < and Ft Ft Σ F. p Σ F Λ]. as for some positive matrix Proof. E F t 4 < follows from E F t 4 < by Assumption 2 and the definition of G t. o prove the second part, we partition the matrix Σ F = lim F t F t into: Σ Σ 2 Σ 2 Σ 22 where Σ = lim F t F t, Σ 22 = lim F 2 t F 2 t, Σ 2 = lim F t F 2 t, and F t is the k subvector of F t that has big breaks in their loadings, F 2 t is the k 2 subvector of F t that doesn t have big breaks in their loadings. By the definition of F t G t we have: and Ft Ft Ft Ft Ft 2 t=τ+ Ft Ft Ft = Ft 2 Ft Ft 2 Ft 2 t=τ+ Ft 2 Ft t=τ+ Ft Ft t=τ+ Ft Ft 2. t=τ+ Ft Ft By Assumption 2, the above matrix converges to Moreover, Σ Σ 2 π Σ Σ F = Σ 2 Σ 22 π Σ 2. π Σ π Σ 2 π Σ Σ Σ 2 0 detσ F = det Σ 2 Σ 22 π Σ 2 = detσ F detπ π Σ > π π Σ because Σ F is positive definite by assumption. his completes the proof.
2 Lemma 2. Γ i < for all i, and Γ Γ Σ Γ as for some positive definite matrix Σ Γ. Proof. his follows directly from Assumptions.a and 3. he following lemmae involve the new errors ɛ t. Let M and M denote some positive constants. Lemma 3. Eɛ it = 0, E ɛ it 8 M for all i and t. Proof. For t =,..., τ, E ɛ it 8 = E e it 8 < M by Assumption 4. For t = τ +,...,, E ɛ it 8 = E e it + η if t E e it 8 + E η if t 2 8 by Loève s inequality. ext, E η i F 2 t 8 η i 8 E F t 8 < by Assumptions.a and 2. hen the result follows. Lemma 4. Eɛ sɛ t / = E i= ɛ is ɛ it = γ s, t, γ s, s M for all s, and s= γ s, t2 M for all t and. Proof. γs, t = Eɛ is ɛ it i= = Ee is + η ig 2 see it + η ig 2 t i= = [ Eeis e it + Eη ig 2 sη ig 2 t ] i= Ee is e it + E η i 2E s G2 η i G 2 2. t i= Since i= Ee is e it = γ s, t by Assumption 4, and E η i 2 G2 t ηi 2 E F t 2 = O for all t by Assumptions.b and 2, we have γ s, t γ s, t + O. hen i= γ s, s γ s, s + O M 2
3 by Assumption 4. Moreover, γs, t 2 s= = s= s= γ s, t + O 2 γ s, t 2 + O M + O M by Assumption 4. hus, the proof is complete. Lemma 5. Eɛ it ɛ jt = τ ij,t with τ ij,t τ ij for some τ ij and for all t; and i= j= τ ij M. Proof. By Assumption 4, τ ij,t τ ij for some τ ij and all t, where τ ij,t = Ee it e jt. hen: for all t. herefore by Assumption 4. τ ij,t = Eɛ it ɛ jt = Ee it + η ig 2 t e jt + η jg 2 t Ee it e jt + E η i 2E s G2 η i G 2 t τ ij + O τij τ ij + O i= j= i= j= M + O Lemma 6. Eɛ it ɛ js = τ ij,ts and i= j= s= τ ij,ts M. M Proof. By Assumption 4, i= j= s= τ ij,ts M, where Ee it e js = τ ij,ts. hen: Eɛ it ɛ js = τ ij,ts = Ee it e js + Eη ig 2 t η jg 2 s = τ ij,ts + Eη ig 2 t η jg 2 s 2 3
4 and we have τij,ts τ ij,ts + Eη ig 2 t η jg 2 s i= j= i= j= i= j= M + O M following the same arguments as above. Lemma 7. For every t, s, E /2 i= [ɛ is ɛ it Eɛ is ɛ it ] 4 M. Proof. Since ɛ it = e it + η i G2 t, we have: ɛ it ɛ is Eɛ it ɛ is = e it e is Ee it e is + e it η ig 2 s + e is η ig 2 t + η ig 2 t η ig 2 s Eη ig 2 t η ig 2 s. Since E e it η i G 2 s 4 η i 4 E e it 4 E F t 4 = O p 2 2, and E η i G2 t η i G2 s 4 η i 8 E F t 4 = O p 4 4, the result follows from Loève s inequality and that E /2 [e is e it Ee is e it ] 4 M. i= Lemma 8. E i= F t ɛ it 2 M. Proof. By the definition of ɛ it we have: E i= Ft 2 ɛ it E i= Ft 2 e it + E i= Ft η ig 2 t 2 then by the definition of F t and G 2 t, F t e it 2 = F t e it 2 + Ft 2 e it, t=τ+ Ft η ig 2 t 2 = First, by Assumption 4 we have t=τ+ F t η if t t=τ+ F t η if 2 t 2. E i= F t e it 2 M. 4
5 Second, and = = E r k= r t=τ+ F t η if 2 t 2 E t=τ+ 2 F kt η if t 2 p= t=τ+ s=τ+ E F kt F ks η if t η if s E F kt F ks η if t 2 η if s 2 η i 2 E F t 4 = O, so we have E t=τ+ F t η i F t 2 2 = O/. he result then follows by noting that t=τ+ Ft 2 e it t=τ+ 2 F t e it and t=τ+ Ft η i F t 2 2 t=τ+ F t η i F t 2 2. As mentioned before, once it has been shown that the new factors: Ft, the new loadings: Γ and the new errors: ɛ t all satisfy the necessary conditions of B 2002, Propositions and 2 just follow directly from their heorems and 2, with r replaced by r + k and F t replaced by Ft. A.2: Proof of heorem We only derive the limiting distributions for the two versions of the LM test, since the proof for the Wald tests is very similar. Let ˆF t define the r vector of estimated factors. Under the null: k = 0, when r = r we have ˆF t = DF t + o p. Let D i denote the ith row of D, and D j denote the jth column of D. Define ˆF t = DF t, and ˆF kt = D k F t as the kth element of ˆFt. Let ˆF t be the first element of ˆFt, and ˆF t = [ ˆF 2t,..., ˆF rt ], while ˆF t and ˆF t can be defined in the same way. ote that ˆF t depends on and. For simplicity, let π denote [ π]. ote that under H 0, we allow for the existence of small breaks, so that the model can be written as X it = α i F t + e it + η i G 2 t. However, since η i G 2 t is O p / by Assumption, we can use similar methods as in Appendix A. to show that an error term of this order 5
6 can be ignored and that the asymptotic properties of ˆF t will not be affected See Remark 5 of Bai, herefore, for simplicity in the presentation below, we eliminate the last term and consider instead the model X it = α i F t + e it in the following lemmae 9 to 3 required to prove Lemma 4 which is the key one in the proof of heorem. Lemma 9. sup π [0,] π Proof. Following Bai 2003 we have: ˆF t ˆF t F t = O p δ 2,. π ˆF t ˆF t F t = 2 π s= π + 2 ˆF s F tξ st s= = I + II + III + IV π ˆF s F tγ s, t + 2 s= π ˆF s F tζ st + 2 s= ˆF s F tκ st where ζ st = e se t γ s, t. κ st = F sa e t /. ξ st = F ta e s /. First, note that: π I = 2 ˆF s DF s F tγ s, t + 2 π D F s F tγ s, t. s= s= Consider the first part of the right hand side, we have π 2 ˆF s DF s F tγ s, t s= = 2 ˆF π s DF s F tγ s, t s= /2 ˆFs DF s 2 π F t 2 s= π γ s, t 2. s= s= ˆFs DF s 2 is Op δ, 2 by heorem of B 2002, sup π [0,] F t 2 = Op by Assumption 2, and sup π [0,] π F t 2 s= π γ s, t 2 s= γ s, t 2 = 6
7 O p by Lemma i of B herefore: sup 2 π π [0,] s= For the second part, note that: ˆF s DF s F tγ s, t = O p δ, /2. sup 2 π D F s F tγ s, t π [0,] s= 2 D F s F t γ s, t s= and E s= s= F s F t γ s, t E F t 2 γ s, t = E F t 2 γ s, t M s= by Assumptions 2 and 4, so the second part is O p given that D is O p. herefore, we have ext, II can be written as: Similarly, we have sup I = O p. A. π [0,] δ, π II = 2 ˆF s DF s F tζ st + 2 π D F s F tζ st. s= s= sup 2 π ˆF s DF s F tζ st π [0,] s= sup ˆFs DF s 2 π F t 2 π π [0,] s= 2 ζst 2 s= ˆFs DF s 2 F t 2 s= 2 ζst 2 s= = O p δ, 7
8 because E ζ st 2 = E /2 [e it e is Ee it e is ] 2 = O i= by Assumption 4. As for the second term of II, we have: where 2 π D F s F tζ st = π s= D q t F t q t = Since E q t 2 M by Assumption 4, we have = [e it e is Ee it e is ]F s. s= i= sup 2 π D F s F tζ st π [0,] s= sup π D q t F t π [0,] D sup π π [0,] O p = O p. q t 2 q t 2 π F t 2 F t 2 hen it follows that Regarding III, it can be written as: sup II = O p. A.2 π [0,] δ, π III = 2 ˆF s DF s F tκ st + 2 π D F s F tκ st s= s= and the second part on the right hand side can be written as D F s F s s= π α i F te it. i= 8
9 herefore: sup 2 π D F s F tκ st π [0,] s= D F s F s sup s= π [0,] = O p π α i F te it i= by Assumption 8. As for the first part on the right hand side of III, we have sup 2 π ˆF s DF s F tκ st π [0,] s= ˆFs DF s 2 sup π F t s= π [0,] κ st 2 s= sup F π [0,] π s s= i= O p δ, 2 F s sup π s= π [0,] i= = O p δ, = O p δ, by Assumption 8. hus, It can also be proved in the similar way that Finally we have: sup π [0,] π ˆF t ˆF t F t α i F t e it 2 α i F t e it sup III = O p. A.3 π [0,] sup IV = O p. A.4 π [0,] sup π [0,] I + sup II + sup π [0,] = O p δ, + O p δ, + O p = O p δ 2, π [0,]. 2 III + sup IV π [0,] 9
10 Lemma 0. Proof. ote that: sup π [0,] π ˆF t ˆF t π ˆF t ˆF t = Op δ 2,. = = = π π π π ˆF t ˆF t ˆF t ˆF t π π ˆF t ˆF t ˆF t ˆF t F td + DF t F td π ˆF t DF t F td ˆF t DF t ˆF t DF t + π D F t ˆF t DF t + π ˆF t DF t F td. hus, sup π [0,] sup π [0,] π π ˆF t ˆF t π ˆF t ˆF t π [0,] ˆF t DF t ˆF t DF t + 2 D sup π [0,] ˆFt DF t D sup π since ˆFt DF t 2 = Op δ, 2 and sup π [0,] Lemma 9, the proof is complete. ˆF t DF t F t π π ˆF t DF t F t ˆF t DF t F t is O p δ, 2 by he next two lemmae follow from Lemma 0 and Assumption 6: Lemma. sup π [0,] π Proof. See Lemma 0 and Assumption 6. ˆF t ˆFt π ˆF t ˆFt = op. Lemma 2. ˆF t ˆF t = op. Proof. By construction we have ˆF t ˆF t = 0, and then the result follows from Lemma. 0
11 Let denote weak convergence. D, F t, F t, F t and S are defined as in the paper see Page 2. Similarly, let Di denote the ith row of D, and D j denote the jth column of D. hen: Lemma 3. π F t F t EF t F t S /2 W r π for π [0, ], where W r is a r vector of independent Brownian motions on [0, ]. Proof. F t F t is stationary and ergodic because F t is stationary and ergodic by Assumption 7. First, we show that {F kt F t EF kt F t, Ω t } is an adapted mixingale of size for k = 2,..., r. By definition, we have F kt F t = D k F td F t = r p= D kp F pt rp= D p F pt = rh= rp= D kp D h F ptf ht, and F kt F t EF kt F t = r h= rp= D kp D h F ptf ht EF pt F ht = rh= rp= D kp D h Y hp,t. hus: = E E 2 F kt F t EF kt F t Ω t m r r 2 E Dkp D h EY hp,t Ω t m h= p= r r DkpD h h= p= r r h= p= c hp t γm hp r 2 max c hp t maxγm hp E EY hp,t Ω t m 2 since maxγ hp m is Om δ for some δ > 0 by Assumption 7, we conclude that {F kt F t EF kt F t, Ω t } is an adapted mixingale of size for k = 2,..., r. ext, we prove the weak convergence using the Crame-Rao device. Define where a R r, and a a =. ote that z t = a S /2 F t F t EF t F t z t = r ã k [F kt F t EF kt F t ] k=2
12 where ã k is the k th element of a S /2. Ez 2 t r 2 2 E ã k [F kt F t EF kt F t ] k=2 r k=2 E F kt F t 2 EF kt F t 2 2 M because E F t 4 < and F kt = D k Ft. Moreover, z t is stationary and ergodic, and we can show {z t, Ω t } is an adapted mixingale sequence of size because: 2 E Ez t Ω t m = r E k=2 ã k E F 2 kt F t EF kt F t Ω t m k=2 r 2 ã k E E F kt F t EF kt F t Ω t m max ã k r c k t γ m. k By the results above we know that γ k m is Om δ for k = 2,..., r. Hence it follows that {z t, Ω t } is an adapted mixingale sequence of size. hen it follows from heorem 7.7 of White 200 that: k=2 π z t = a S /2 π F t F t EF t F t Wπ. Moreover, it can be shown that: a π 2 t= π F t F t EF t F t +a π 0 2 F t F t EF t F t d 0, π2 π a Sa +π 0 a 2Sa 2 by using Corollary 3. of Woodridge and White 988. he proof is completed by using Lemma A.4 of Andrews 993. A.3: More discussions on Remark 9 In Remark 9 of the paper we mention that, although our tests are designed for single break, they should also have power against multiple breaks. o see this, consider the simple example of a FM with one factor and two big breaks: X t = Af t t τ + Bf t τ < t < τ 2 + Df t t τ 2 + e t 2
13 = Ag t + Bh t + Ds t + e t where g t = f t t τ, h t = f t τ < t < τ 2, and s t = f t t τ 2. In view of Proposition 2, Bai and g s 2002 IC will lead to the choice of 3 factors which, when estimated by PCA, implies the following result: ˆf t d d 4 d 7 g t ˆf 2t = d 2 d 5 d 8 h t + o p. ˆf 3t d 3 d 6 d 9 s t hen, by the definition of g t, h t and s t we have: ˆf t d ˆf 2t = d 2 f t + o p for t =,..., τ, ˆf 3t d 3 ˆf t d 4 ˆf 2t = d 5 f t + o p for t = τ,..., τ 2, ˆf 3t d 6 ˆf t d 7 ˆf 2t = d 8 f t + o p for t = τ 2,...,. ˆf 3t d 9 Hence, we can find one vector [p, p 2, p 3 ] which is orthogonal to [d, d 2, d 3 ] and [d 4, d 5, d 6 ], plus another vector [p 4, p 5, p 6 ] which is orthogonal to [d 7, d 8, d 9 ]. It is easy to see that [p, p 2, p 3 ] a[p 4, p 5, p 6 ] for any a 0 otherwise the D matrix will be singular, and thus we can find a breaking relationship between the estimated factors and even use Bai and Perron s 998, 2003 to detect a second break. he simulation results about the power of our tests against multiple breaks are available upon request. 3
14 References Andrews, D ests for parameter instability and structural change with unknown change point. Econometrica 6 4, Bai, J Inferential theory for factor models of large dimensions. Econometrica 7, Bai, J Panel data models with interactive fixed effects. Econometrica 77 4, Bai, J. and S. g Determining the number of factors in approximate factor models. Econometrica 70, Bai, J. and P. Perron 998. Estimating and testing linear models with multiple structural changes. Econometrica 66, Bai, J. and P. Perron Computation and analysis of multiple structural change models. Journal of Applied Econometrics 8, 22. White, H Asymptotic theory for econometricians. Academic Press Orlando, Florida. Wooldridge, J. and H. White 988. Some invariance principles and central limit theorems for dependent heterogeneous processes. Econometric heory 4 2,
More Empirical Process Theory
More Empirical Process heory 4.384 ime Series Analysis, Fall 2008 Recitation by Paul Schrimpf Supplementary to lectures given by Anna Mikusheva October 24, 2008 Recitation 8 More Empirical Process heory
More informationDetermining the Number of Factors When the Number of Factors Can Increase with Sample Size
Determining the umber of Factors When the umber of Factors Can Increase with Sample Size Hongjun Li International School of Economics and Management, Capital University of Economics and Business, Beijing,
More informationA panel unit root test in the presence of cross-section dependence
A panel unit root test in the presence of cross-section dependence Mehdi Hosseinkouchack Department of Economics, Goethe University Frankfurt, Grueneburgplatz, (House of Finance, 60323, Frankfurt am Main,
More informationFederal Reserve Bank of New York Staff Reports
Federal Reserve Bank of New York Staff Reports One-Sided est for an Unknown Breakpoint: heory, Computation, and Application to Monetary heory Arturo Estrella Anthony P. Rodrigues Staff Report no. 232 November
More informationProofs for Large Sample Properties of Generalized Method of Moments Estimators
Proofs for Large Sample Properties of Generalized Method of Moments Estimators Lars Peter Hansen University of Chicago March 8, 2012 1 Introduction Econometrica did not publish many of the proofs in my
More informationEstimation of Structural Breaks in Large Panels with Cross-Sectional Dependence
ISS 440-77X Department of Econometrics and Business Statistics http://business.monash.edu/econometrics-and-business-statistics/research/publications Estimation of Structural Breaks in Large Panels with
More informationPutzer s Algorithm. Norman Lebovitz. September 8, 2016
Putzer s Algorithm Norman Lebovitz September 8, 2016 1 Putzer s algorithm The differential equation dx = Ax, (1) dt where A is an n n matrix of constants, possesses the fundamental matrix solution exp(at),
More informationA DIRECT TEST FOR CROSS-SECTIONAL CORRELATION IN PANEL DATA MODELS
A DIREC ES FOR CROSS-SECIOAL CORRELAIO I PAEL DAA MODELS ED JUHL Abstract. his paper proposes a test for cross-sectional correlation for use in panel data models. I propose a test that uses squares of
More information3. WIENER PROCESS. STOCHASTIC ITÔ INTEGRAL
3. WIENER PROCESS. STOCHASTIC ITÔ INTEGRAL 3.1. Wiener process. Main properties. A Wiener process notation W W t t is named in the honor of Prof. Norbert Wiener; other name is the Brownian motion notation
More informationBreak Point Estimation in Fixed Effects Panel Data
Break Point Estimation in Fixed Effects Panel Data Otilia Boldea, Zhuojiong Gan and Bettina Drepper February 15, 2016 Abstract In this paper, we propose estimating multiple break-points in panel data with
More informationLecture 17 Brownian motion as a Markov process
Lecture 17: Brownian motion as a Markov process 1 of 14 Course: Theory of Probability II Term: Spring 2015 Instructor: Gordan Zitkovic Lecture 17 Brownian motion as a Markov process Brownian motion is
More informationSupplemental Material for KERNEL-BASED INFERENCE IN TIME-VARYING COEFFICIENT COINTEGRATING REGRESSION. September 2017
Supplemental Material for KERNEL-BASED INFERENCE IN TIME-VARYING COEFFICIENT COINTEGRATING REGRESSION By Degui Li, Peter C. B. Phillips, and Jiti Gao September 017 COWLES FOUNDATION DISCUSSION PAPER NO.
More informationINFERENTIAL THEORY FOR FACTOR MODELS OF LARGE DIMENSIONS. Jushan Bai. May 21, 2002
IFEREIAL HEORY FOR FACOR MODELS OF LARGE DIMESIOS Jushan Bai May 2, 22 Abstract his paper develops an inferential theory for factor models of large dimensions. he principal components estimator is considered
More informationLinear Ordinary Differential Equations
MTH.B402; Sect. 1 20180703) 2 Linear Ordinary Differential Equations Preliminaries: Matrix Norms. Denote by M n R) the set of n n matrix with real components, which can be identified the vector space R
More informationA PANIC Attack on Unit Roots and Cointegration. July 31, Preliminary and Incomplete
A PANIC Attack on Unit Roots and Cointegration Jushan Bai Serena Ng July 3, 200 Preliminary and Incomplete Abstract his paper presents a toolkit for Panel Analysis of Non-stationarity in Idiosyncratic
More informationAsymptotic Distributions of Instrumental Variables Statistics with Many Instruments
CHAPTER 6 Asymptotic Distributions of Instrumental Variables Statistics with Many Instruments James H. Stock and Motohiro Yogo ABSTRACT This paper extends Staiger and Stock s (1997) weak instrument asymptotic
More informationProperties of the stress tensor
Appendix C Properties of the stress tensor Some of the basic properties of the stress tensor and traction vector are reviewed in the following. C.1 The traction vector Let us assume that the state of stress
More informationSupplement to Covariance Matrix Estimation for Stationary Time Series
Supplement to Covariance Matrix Estimation for Stationary ime Series Han Xiao and Wei Biao Wu Department of Statistics 5734 S. University Ave Chicago, IL 60637 e-mail: xiao@galton.uchicago.edu e-mail:
More informationOn Perron s Unit Root Tests in the Presence. of an Innovation Variance Break
Applied Mathematical Sciences, Vol. 3, 2009, no. 27, 1341-1360 On Perron s Unit Root ests in the Presence of an Innovation Variance Break Amit Sen Department of Economics, 3800 Victory Parkway Xavier University,
More informationTopological Properties of Operations on Spaces of Continuous Functions and Integrable Functions
Topological Properties of Operations on Spaces of Continuous Functions and Integrable Functions Holly Renaud University of Memphis hrenaud@memphis.edu May 3, 2018 Holly Renaud (UofM) Topological Properties
More informationOnline Appendix. j=1. φ T (ω j ) vec (EI T (ω j ) f θ0 (ω j )). vec (EI T (ω) f θ0 (ω)) = O T β+1/2) = o(1), M 1. M T (s) exp ( isω)
Online Appendix Proof of Lemma A.. he proof uses similar arguments as in Dunsmuir 979), but allowing for weak identification and selecting a subset of frequencies using W ω). It consists of two steps.
More informationBootstrap prediction intervals for factor models
Bootstrap prediction intervals for factor models Sílvia Gonçalves and Benoit Perron Département de sciences économiques, CIREQ and CIRAO, Université de Montréal April, 3 Abstract We propose bootstrap prediction
More informationState-space Model. Eduardo Rossi University of Pavia. November Rossi State-space Model Financial Econometrics / 49
State-space Model Eduardo Rossi University of Pavia November 2013 Rossi State-space Model Financial Econometrics - 2013 1 / 49 Outline 1 Introduction 2 The Kalman filter 3 Forecast errors 4 State smoothing
More informationA NOTE ON SPURIOUS BREAK
Econometric heory, 14, 1998, 663 669+ Printed in the United States of America+ A NOE ON SPURIOUS BREAK JUSHAN BAI Massachusetts Institute of echnology When the disturbances of a regression model follow
More informationA PANIC Attack on Unit Roots and Cointegration
A PANIC Attack on Unit Roots and Cointegration Authors: Jushan Bai, Serena Ng his work is posted on escholarship@bc, Boston College University Libraries. Boston College Working Papers in Economics, 200
More informationZhaoxing Gao and Ruey S Tsay Booth School of Business, University of Chicago. August 23, 2018
Supplementary Material for Structural-Factor Modeling of High-Dimensional Time Series: Another Look at Approximate Factor Models with Diverging Eigenvalues Zhaoxing Gao and Ruey S Tsay Booth School of
More informationComparing Forecast Accuracy of Different Models for Prices of Metal Commodities
Comparing Forecast Accuracy of Different Models for Prices of Metal Commodities João Victor Issler (FGV) and Claudia F. Rodrigues (VALE) August, 2012 J.V. Issler and C.F. Rodrigues () Forecast Models for
More informationSimultaneous change-point and factor analysis for high-dimensional time series
Simultaneous change-point and factor analysis for high-dimensional time series Piotr Fryzlewicz Joint work with Haeran Cho and Matteo Barigozzi (slides courtesy of Haeran) CMStatistics 2017 Department
More informationChapter One. The Calderón-Zygmund Theory I: Ellipticity
Chapter One The Calderón-Zygmund Theory I: Ellipticity Our story begins with a classical situation: convolution with homogeneous, Calderón- Zygmund ( kernels on R n. Let S n 1 R n denote the unit sphere
More informationESTIMATING AND TESTING HIGH DIMENSIONAL FACTOR MODELS WITH MULTIPLE STRUCTURAL CHANGES. Chihwa Kao Syracuse University.
ESIMAING AND ESING HIGH DIMENSIONAL FACOR MODELS WIH MULIPLE SRUCURAL CHANGES Badi H. Baltagi Syracuse University Chihwa Kao Syracuse University February 0, 06 Fa Wang Syracuse University Abstract his
More informationObserver design for a general class of triangular systems
1st International Symposium on Mathematical Theory of Networks and Systems July 7-11, 014. Observer design for a general class of triangular systems Dimitris Boskos 1 John Tsinias Abstract The paper deals
More informationTESTING SERIAL CORRELATION IN SEMIPARAMETRIC VARYING COEFFICIENT PARTIALLY LINEAR ERRORS-IN-VARIABLES MODEL
Jrl Syst Sci & Complexity (2009) 22: 483 494 TESTIG SERIAL CORRELATIO I SEMIPARAMETRIC VARYIG COEFFICIET PARTIALLY LIEAR ERRORS-I-VARIABLES MODEL Xuemei HU Feng LIU Zhizhong WAG Received: 19 September
More information: Error Correcting Codes. December 2017 Lecture 10
0368307: Error Correcting Codes. December 017 Lecture 10 The Linear-Programming Bound Amnon Ta-Shma and Dean Doron 1 The LP-bound We will prove the Linear-Programming bound due to [, 1], which gives an
More informationDepartment of Economics, Boston University. Department of Economics, Boston University
Supplementary Material Supplement to Identification and frequency domain quasi-maximum likelihood estimation of linearized dynamic stochastic general equilibrium models : Appendix (Quantitative Economics,
More informationUnit Root Testing in Heteroskedastic Panels using the Cauchy Estimator
Unit Root esting in Heteroskedastic Panels using the Cauchy Estimator Matei Demetrescu Christoph Hanck Preliminary version: June 25, 2010 Abstract he so-called Cauchy estimator uses the sign of the first
More informationOn an Additive Semigraphoid Model for Statistical Networks With Application to Nov Pathway 25, 2016 Analysis -1 Bing / 38Li,
On an Additive Semigraphoid Model for Statistical Networks With Application to Pathway Analysis - Bing Li, Hyunho Chun & Hongyu Zhao Kim Youngrae SNU Stat. Multivariate Lab Nov 25, 2016 On an Additive
More informationAn estimate of the long-run covariance matrix, Ω, is necessary to calculate asymptotic
Chapter 6 ESTIMATION OF THE LONG-RUN COVARIANCE MATRIX An estimate of the long-run covariance matrix, Ω, is necessary to calculate asymptotic standard errors for the OLS and linear IV estimators presented
More informationTitle. Description. Remarks and examples. stata.com. stata.com. Variable notation. methods and formulas for sem Methods and formulas for sem
Title stata.com methods and formulas for sem Methods and formulas for sem Description Remarks and examples References Also see Description The methods and formulas for the sem commands are presented below.
More informationI forgot to mention last time: in the Ito formula for two standard processes, putting
I forgot to mention last time: in the Ito formula for two standard processes, putting dx t = a t dt + b t db t dy t = α t dt + β t db t, and taking f(x, y = xy, one has f x = y, f y = x, and f xx = f yy
More informationAsymptotic Properties of an Approximate Maximum Likelihood Estimator for Stochastic PDEs
Asymptotic Properties of an Approximate Maximum Likelihood Estimator for Stochastic PDEs M. Huebner S. Lototsky B.L. Rozovskii In: Yu. M. Kabanov, B. L. Rozovskii, and A. N. Shiryaev editors, Statistics
More informationThe Almost-Sure Ergodic Theorem
Chapter 24 The Almost-Sure Ergodic Theorem This chapter proves Birkhoff s ergodic theorem, on the almostsure convergence of time averages to expectations, under the assumption that the dynamics are asymptotically
More informationDarmstadt Discussion Papers in Economics
Darmstadt Discussion Papers in Economics The Effect of Linear Time Trends on Cointegration Testing in Single Equations Uwe Hassler Nr. 111 Arbeitspapiere des Instituts für Volkswirtschaftslehre Technische
More informationOutline of the course
School of Mathematical Sciences PURE MTH 3022 Geometry of Surfaces III, Semester 2, 20 Outline of the course Contents. Review 2. Differentiation in R n. 3 2.. Functions of class C k 4 2.2. Mean Value Theorem
More informationLecture 4: Numerical solution of ordinary differential equations
Lecture 4: Numerical solution of ordinary differential equations Department of Mathematics, ETH Zürich General explicit one-step method: Consistency; Stability; Convergence. High-order methods: Taylor
More informationBrownian Motion. 1 Definition Brownian Motion Wiener measure... 3
Brownian Motion Contents 1 Definition 2 1.1 Brownian Motion................................. 2 1.2 Wiener measure.................................. 3 2 Construction 4 2.1 Gaussian process.................................
More informationIntroduction Benchmark model Belief-based model Empirical analysis Summary. Riot Networks. Lachlan Deer Michael D. König Fernando Vega-Redondo
Riot Networks Lachlan Deer Michael D. König Fernando Vega-Redondo University of Zurich University of Zurich Bocconi University June 7, 2018 Deer & König &Vega-Redondo Riot Networks June 7, 2018 1 / 23
More informationAsymptotically unbiased estimation of auto-covariances and auto-correlations with long panel data
Asymptotically unbiased estimation of auto-covariances and auto-correlations with long panel data Ryo Oui February 2, 2007 Incomplete and Preliminary Abstract Many economic variables are correlated over
More informationCentral Bank of Chile October 29-31, 2013 Bruce Hansen (University of Wisconsin) Structural Breaks October 29-31, / 91. Bruce E.
Forecasting Lecture 3 Structural Breaks Central Bank of Chile October 29-31, 2013 Bruce Hansen (University of Wisconsin) Structural Breaks October 29-31, 2013 1 / 91 Bruce E. Hansen Organization Detection
More informationThe heat equation in time dependent domains with Neumann boundary conditions
The heat equation in time dependent domains with Neumann boundary conditions Chris Burdzy Zhen-Qing Chen John Sylvester Abstract We study the heat equation in domains in R n with insulated fast moving
More informationN-CONSISTENT SEMIPARAMETRIC REGRESSION: UNIT ROOT TESTS WITH NONLINEARITIES. 1. Introduction
-COSISTET SEMIPARAMETRIC REGRESSIO: UIT ROOT TESTS WITH OLIEARITIES TED JUHL AD ZHIJIE XIAO Abstract. We develop unit root tests using additional time series as suggested in Hansen 995). However, we allow
More informationZero-sum square matrices
Zero-sum square matrices Paul Balister Yair Caro Cecil Rousseau Raphael Yuster Abstract Let A be a matrix over the integers, and let p be a positive integer. A submatrix B of A is zero-sum mod p if the
More informationThis chapter reviews properties of regression estimators and test statistics based on
Chapter 12 COINTEGRATING AND SPURIOUS REGRESSIONS This chapter reviews properties of regression estimators and test statistics based on the estimators when the regressors and regressant are difference
More informationLimiting behaviour of large Frobenius numbers. V.I. Arnold
Limiting behaviour of large Frobenius numbers by J. Bourgain and Ya. G. Sinai 2 Dedicated to V.I. Arnold on the occasion of his 70 th birthday School of Mathematics, Institute for Advanced Study, Princeton,
More informationAssumptions for the theoretical properties of the
Statistica Sinica: Supplement IMPUTED FACTOR REGRESSION FOR HIGH- DIMENSIONAL BLOCK-WISE MISSING DATA Yanqing Zhang, Niansheng Tang and Annie Qu Yunnan University and University of Illinois at Urbana-Champaign
More informationMarkov Chains and Stochastic Sampling
Part I Markov Chains and Stochastic Sampling 1 Markov Chains and Random Walks on Graphs 1.1 Structure of Finite Markov Chains We shall only consider Markov chains with a finite, but usually very large,
More informationGeneralized Method of Moments: I. Chapter 9, R. Davidson and J.G. MacKinnon, Econometric Theory and Methods, 2004, Oxford.
Generalized Method of Moments: I References Chapter 9, R. Davidson and J.G. MacKinnon, Econometric heory and Methods, 2004, Oxford. Chapter 5, B. E. Hansen, Econometrics, 2006. http://www.ssc.wisc.edu/~bhansen/notes/notes.htm
More informationLocation Multiplicative Error Model. Asymptotic Inference and Empirical Analysis
: Asymptotic Inference and Empirical Analysis Qian Li Department of Mathematics and Statistics University of Missouri-Kansas City ql35d@mail.umkc.edu October 29, 2015 Outline of Topics Introduction GARCH
More informationQuick Review on Linear Multiple Regression
Quick Review on Linear Multiple Regression Mei-Yuan Chen Department of Finance National Chung Hsing University March 6, 2007 Introduction for Conditional Mean Modeling Suppose random variables Y, X 1,
More informationSemiparametric Trending Panel Data Models with Cross-Sectional Dependence
he University of Adelaide School of Economics Research Paper No. 200-0 May 200 Semiparametric rending Panel Data Models with Cross-Sectional Dependence Jia Chen, Jiti Gao, and Degui Li Semiparametric rending
More informationOn the Local Convergence of an Iterative Approach for Inverse Singular Value Problems
On the Local Convergence of an Iterative Approach for Inverse Singular Value Problems Zheng-jian Bai Benedetta Morini Shu-fang Xu Abstract The purpose of this paper is to provide the convergence theory
More informationDerivatives of Harmonic Bergman and Bloch Functions on the Ball
Journal of Mathematical Analysis and Applications 26, 1 123 (21) doi:1.16/jmaa.2.7438, available online at http://www.idealibrary.com on Derivatives of Harmonic ergman and loch Functions on the all oo
More informationState-space Model. Eduardo Rossi University of Pavia. November Rossi State-space Model Fin. Econometrics / 53
State-space Model Eduardo Rossi University of Pavia November 2014 Rossi State-space Model Fin. Econometrics - 2014 1 / 53 Outline 1 Motivation 2 Introduction 3 The Kalman filter 4 Forecast errors 5 State
More informationCUSUM TEST FOR PARAMETER CHANGE IN TIME SERIES MODELS. Sangyeol Lee
CUSUM TEST FOR PARAMETER CHANGE IN TIME SERIES MODELS Sangyeol Lee 1 Contents 1. Introduction of the CUSUM test 2. Test for variance change in AR(p) model 3. Test for Parameter Change in Regression Models
More informationarxiv: v1 [econ.em] 19 Feb 2019
Estimation and Inference for Synthetic Control Methods with Spillover Effects arxiv:1902.07343v1 [econ.em] 19 Feb 2019 Jianfei Cao Connor Dowd February 21, 2019 Abstract he synthetic control method is
More informationEstimation of Time-invariant Effects in Static Panel Data Models
Estimation of Time-invariant Effects in Static Panel Data Models M. Hashem Pesaran University of Southern California, and Trinity College, Cambridge Qiankun Zhou University of Southern California September
More informationOn singular values distribution of a matrix large auto-covariance in the ultra-dimensional regime. Title
itle On singular values distribution of a matrix large auto-covariance in the ultra-dimensional regime Authors Wang, Q; Yao, JJ Citation Random Matrices: heory and Applications, 205, v. 4, p. article no.
More information7. MULTIVARATE STATIONARY PROCESSES
7. MULTIVARATE STATIONARY PROCESSES 1 1 Some Preliminary Definitions and Concepts Random Vector: A vector X = (X 1,..., X n ) whose components are scalar-valued random variables on the same probability
More informationMultivariate Analysis and Likelihood Inference
Multivariate Analysis and Likelihood Inference Outline 1 Joint Distribution of Random Variables 2 Principal Component Analysis (PCA) 3 Multivariate Normal Distribution 4 Likelihood Inference Joint density
More informationPersistence Heterogeneity Testing in Panels with Interactive Fixed Effects. Yunus Emre Ergemen and Carlos Velasco. CREATES Research Paper
Persistence Heterogeneity esting in Panels with Interactive Fixed Effects Yunus Emre Ergemen and Carlos Velasco CREAES Research Paper 8- Department of Economics and Business Economics Aarhus University
More informationLecture 7. 1 Notations. Tel Aviv University Spring 2011
Random Walks and Brownian Motion Tel Aviv University Spring 2011 Lecture date: Apr 11, 2011 Lecture 7 Instructor: Ron Peled Scribe: Yoav Ram The following lecture (and the next one) will be an introduction
More informationarxiv: v1 [math.pr] 4 Nov 2016
Perturbations of continuous-time Markov chains arxiv:1611.1254v1 [math.pr 4 Nov 216 Pei-Sen Li School of Mathematical Sciences, Beijing Normal University, Beijing 1875, China E-ma: peisenli@ma.bnu.edu.cn
More informationMissing dependent variables in panel data models
Missing dependent variables in panel data models Jason Abrevaya Abstract This paper considers estimation of a fixed-effects model in which the dependent variable may be missing. For cross-sectional units
More informationFluid Models of Parallel Service Systems under FCFS
Fluid Models of Parallel Service Systems under FCFS Hanqin Zhang Business School, National University of Singapore Joint work with Yuval Nov and Gideon Weiss from The University of Haifa, Israel Queueing
More informationCALCULATION METHOD FOR NONLINEAR DYNAMIC LEAST-ABSOLUTE DEVIATIONS ESTIMATOR
J. Japan Statist. Soc. Vol. 3 No. 200 39 5 CALCULAION MEHOD FOR NONLINEAR DYNAMIC LEAS-ABSOLUE DEVIAIONS ESIMAOR Kohtaro Hitomi * and Masato Kagihara ** In a nonlinear dynamic model, the consistency and
More information1 Math 241A-B Homework Problem List for F2015 and W2016
1 Math 241A-B Homework Problem List for F2015 W2016 1.1 Homework 1. Due Wednesday, October 7, 2015 Notation 1.1 Let U be any set, g be a positive function on U, Y be a normed space. For any f : U Y let
More informationLecture 9. d N(0, 1). Now we fix n and think of a SRW on [0,1]. We take the k th step at time k n. and our increments are ± 1
Random Walks and Brownian Motion Tel Aviv University Spring 011 Lecture date: May 0, 011 Lecture 9 Instructor: Ron Peled Scribe: Jonathan Hermon In today s lecture we present the Brownian motion (BM).
More informationDYNAMIC PANEL GMM WITH NEAR UNITY. Peter C. B. Phillips. December 2014 COWLES FOUNDATION DISCUSSION PAPER NO. 1962
DYNAMIC PANEL GMM WIH NEAR UNIY By Peter C. B. Phillips December 04 COWLES FOUNDAION DISCUSSION PAPER NO. 96 COWLES FOUNDAION FOR RESEARCH IN ECONOMICS YALE UNIVERSIY Box 088 New Haven, Connecticut 0650-88
More informationLecture 8 Inequality Testing and Moment Inequality Models
Lecture 8 Inequality Testing and Moment Inequality Models Inequality Testing In the previous lecture, we discussed how to test the nonlinear hypothesis H 0 : h(θ 0 ) 0 when the sample information comes
More informationCONVERGENCE BEHAVIOUR OF SOLUTIONS TO DELAY CELLULAR NEURAL NETWORKS WITH NON-PERIODIC COEFFICIENTS
Electronic Journal of Differential Equations, Vol. 2007(2007), No. 46, pp. 1 7. ISSN: 1072-6691. URL: http://ejde.math.txstate.edu or http://ejde.math.unt.edu ftp ejde.math.txstate.edu (login: ftp) CONVERGENCE
More informationL2 gains and system approximation quality 1
Massachusetts Institute of Technology Department of Electrical Engineering and Computer Science 6.242, Fall 24: MODEL REDUCTION L2 gains and system approximation quality 1 This lecture discusses the utility
More informationMATH 205C: STATIONARY PHASE LEMMA
MATH 205C: STATIONARY PHASE LEMMA For ω, consider an integral of the form I(ω) = e iωf(x) u(x) dx, where u Cc (R n ) complex valued, with support in a compact set K, and f C (R n ) real valued. Thus, I(ω)
More informationStationary mean-field games Diogo A. Gomes
Stationary mean-field games Diogo A. Gomes We consider is the periodic stationary MFG, { ɛ u + Du 2 2 + V (x) = g(m) + H ɛ m div(mdu) = 0, (1) where the unknowns are u : T d R, m : T d R, with m 0 and
More informationUniformly Uniformly-ergodic Markov chains and BSDEs
Uniformly Uniformly-ergodic Markov chains and BSDEs Samuel N. Cohen Mathematical Institute, University of Oxford (Based on joint work with Ying Hu, Robert Elliott, Lukas Szpruch) Centre Henri Lebesgue,
More informationSelecting the Correct Number of Factors in Approximate Factor Models: The Large Panel Case with Group Bridge Estimators
Selecting the Correct umber of Factors in Approximate Factor Models: The Large Panel Case with Group Bridge Estimators Mehmet Caner Xu Han orth Carolina State University City University of Hong Kong December
More informationEstimation and Testing for Common Cycles
Estimation and esting for Common Cycles Anders Warne February 27, 2008 Abstract: his note discusses estimation and testing for the presence of common cycles in cointegrated vector autoregressions A simple
More informationMultivariate modelling of long memory processes with common components
Multivariate modelling of long memory processes with common components Claudio Morana University of Piemonte Orientale, International Centre for Economic Research (ICER), and Michigan State University
More informationA CLT FOR MULTI-DIMENSIONAL MARTINGALE DIFFERENCES IN A LEXICOGRAPHIC ORDER GUY COHEN. Dedicated to the memory of Mikhail Gordin
A CLT FOR MULTI-DIMENSIONAL MARTINGALE DIFFERENCES IN A LEXICOGRAPHIC ORDER GUY COHEN Dedicated to the memory of Mikhail Gordin Abstract. We prove a central limit theorem for a square-integrable ergodic
More informationA property of the Hodrick-Prescott filter and its application
A property of the Hodrick-Prescott filter and its application (Job Market Paper #2 ) Neslihan Sakarya Robert M. de Jong November 14, 2016 Abstract This paper explores a remarkable property of the Hodrick-Prescott
More informationSPECTRAL THEOREM FOR COMPACT SELF-ADJOINT OPERATORS
SPECTRAL THEOREM FOR COMPACT SELF-ADJOINT OPERATORS G. RAMESH Contents Introduction 1 1. Bounded Operators 1 1.3. Examples 3 2. Compact Operators 5 2.1. Properties 6 3. The Spectral Theorem 9 3.3. Self-adjoint
More informationThe Functional Central Limit Theorem and Testing for Time Varying Parameters
NBER Summer Institute Minicourse What s New in Econometrics: ime Series Lecture : July 4, 008 he Functional Central Limit heorem and esting for ime Varying Parameters Lecture -, July, 008 Outline. FCL.
More informationComparison of Sums of Independent Identically Distributed Random Variables
Comparison of Sums of Independent Identically Distributed Random Variables S.J. Montgomery-Smith* Department of Mathematics, University of Missouri, Columbia, MO 65211. ABSTRACT: Let S k be the k-th partial
More informationDynamic time series binary choice
Dynamic time series binary choice Robert M. de Jong Tiemen M. Woutersen April 29, 2005 Abstract This paper considers dynamic time series binary choice models. It proves near epoch dependence and strong
More informationZeros and zero dynamics
CHAPTER 4 Zeros and zero dynamics 41 Zero dynamics for SISO systems Consider a linear system defined by a strictly proper scalar transfer function that does not have any common zero and pole: g(s) =α p(s)
More informationON BOUNDEDNESS OF THE CURVE GIVEN BY ITS CURVATURE AND TORSION
ON BOUNDEDNESS OF THE CURVE GIVEN BY ITS CURVATURE AND TORSION OLEG ZUBELEVICH DEPT. OF THEORETICAL MECHANICS, MECHANICS AND MATHEMATICS FACULTY, M. V. LOMONOSOV MOSCOW STATE UNIVERSITY RUSSIA, 9899, MOSCOW,
More informationSpecification Test on Mixed Logit Models
Specification est on Mixed Logit Models Jinyong Hahn UCLA Jerry Hausman MI December 1, 217 Josh Lustig CRA Abstract his paper proposes a specification test of the mixed logit models, by generalizing Hausman
More information1 The Glivenko-Cantelli Theorem
1 The Glivenko-Cantelli Theorem Let X i, i = 1,..., n be an i.i.d. sequence of random variables with distribution function F on R. The empirical distribution function is the function of x defined by ˆF
More informationTechnical Appendix for: When Promotions Meet Operations: Cross-Selling and Its Effect on Call-Center Performance
Technical Appendix for: When Promotions Meet Operations: Cross-Selling and Its Effect on Call-Center Performance In this technical appendix we provide proofs for the various results stated in the manuscript
More informationHopping transport in disordered solids
Hopping transport in disordered solids Dominique Spehner Institut Fourier, Grenoble, France Workshop on Quantum Transport, Lexington, March 17, 2006 p. 1 Outline of the talk Hopping transport Models for
More informationA CONVERGENCE RESULT FOR MATRIX RICCATI DIFFERENTIAL EQUATIONS ASSOCIATED WITH M-MATRICES
A CONVERGENCE RESULT FOR MATRX RCCAT DFFERENTAL EQUATONS ASSOCATED WTH M-MATRCES CHUN-HUA GUO AND BO YU Abstract. The initial value problem for a matrix Riccati differential equation associated with an
More informationin Mobile In-App Advertising
Online Appendices to Not Just a Fad: Optimal Sequencing in Mobile In-App Advertising Appendix A: Proofs of echnical Results Proof of heorem : For ease of exposition, we will use a concept of dummy ads
More information