FIXED-b ASYMPTOTICS FOR BLOCKWISE EMPIRICAL LIKELIHOOD
|
|
- Drusilla O’Neal’
- 5 years ago
- Views:
Transcription
1 Statistica Sinica 24 (214), - doi: FIXED-b ASYMPTOTICS FOR BLOCKWISE EMPIRICAL LIKELIHOOD Xianyang Zhang and Xiaofeng Shao University of Missouri-Columbia and University of Illinois at Urbana-Champaign Abstract: We describe an extension of the fixed-b approach introduced by Kiefer and Vogelsang (25) to the empirical likelihood estimation framework. Under fixed-b asymptotics, the empirical likelihood ratio statistic evaluated at the true parameter converges to a nonstandard yet pivotal limiting distribution that can be approximated numerically. The impact of the bandwidth parameter and kernel choice is reflected in the fixed-b limiting distribution. Compared to the χ 2 -based inference procedure used by Kitamura (1997) and Smith (211), the fixed-b approach provides a better approximation to the finite sample distribution of the empirical likelihood ratio statistic. Correspondingly, as shown in our simulation studies, the confidence region based on the fixed-b approach has more accurate coverage than its traditional counterpart. Key words and phrases: Blocking, empirical likelihood, fixed-b asymptotics, time series. 1. Introduction Empirical likelihood (EL) (Owen (1988, 199)) is a nonparametric technique for conducting inference for parameters in nonparametric settings. EL has been studied extensively in the statistics and econometrics literature (see Owen (21), Kitamura (26), and Chen and Van Keilegom (29) for comprehensive reviews). One striking property of EL is the nonparametric version of Wilks theorem that states that the EL ratio statistic evaluated at the true parameter converges to a χ 2 liming distribution. This property was first demonstrated for the mean parameter by Owen (199) and was further extended to the estimating equation framework by Qin and Lawless (1994). However, Wilks phenomenon fails to hold for stationary time series because the dependence within the observations is not taken into account in EL. Kitamura (1997) proposed blockwise empirical likelihood (BEL), which is able to accommodate the dependence of the data, and Wilks theorem continues to hold for the BEL ratio statistic under suitable weak dependence assumptions. The BEL can be viewed as a special case of the generalized empirical likelihood (GEL) with smoothed moment conditions (Smith (211)).
2 2 XIANYANG ZHANG AND XIAOFENG SHAO The performance of BEL and its variations (Nordman (29) and Smith (211)) can depend crucially on the choice of the bandwidth parameter for which no sound guidance is available. Kiefer and Vogelsang (25) proposed the socalled fixed-b asymptotic theory in the heteroscedasticity-autocorrelation robust (HAR) testing context. It was found that the asymptotic distribution obtained by treating the bandwidth as a fixed proportion (say b) of the sample size provides a better approximation to the sampling distribution of the studentized test statistic than the traditional χ 2 -based approximation. See Jansson (24), Sun, Phillips, and Jin (28), and Zhang and Shao (213) for rigorous theoretical justifications. The fixed-b approach has the advantage of accounting for the effect of the bandwidth and the kernel, as different bandwidth parameters and kernels correspond to different limiting (null) distributions (also see Shao and Politis (213) for a recent extension to the subsampling and block bootstrap context). The main thrust of the present paper is the development of a new asymptotic theory in the BEL estimation framework made possible by the fixed-b approach. We consider the problem in the moment condition model (Qin and Lawless (1994) and Smith (211)) that is a fairly general framework used by both statisticians and econometricians. Under the fixed-b asymptotic framework, we show that the asymptotic null distribution of the EL ratio statistic evaluated at the true parameter is nonstandard yet pivotal, and that it can be approximated numerically. It is interesting to note that the fixed-b limiting distribution coincides with the χ 2 distribution as b gets close to zero. We also illustrate the idea in the GEL estimation framework and demonstrate the usefulness of the fixed-b approach through simulation studies. For notation, let D[, 1] be the space of functions on [, 1] which are rightcontinuous and have left limits, endowed with the Skorokhod topology (see Billingsley (1999)). Weak convergence in D[, 1], or more generally in the R m - valued function space D m [, 1], is denoted by, where m N. Convergence in probability and convergence in distribution are denoted by p and d respectively. Let C[, 1] be the space of continuous functions on [, 1]. Denote by a the integer part of a R. 2. Methodology 2.1. Empirical likelihood Suppose we are interested in the inference of a p-dimensional parameter vector θ that is identified by a set of moment conditions. Denote by θ the true parameter of θ, an interior point of a compact parameter space Θ R p. Let {y t } n t=1 be a sequence of Rl -valued stationary time series and assume the moment conditions E[f(y t, θ )] =, t = 1, 2,..., n, (2.1)
3 FIXED-b ASYMPTOTICS FOR BLOCKWISE EMPIRICAL LIKELIHOOD 3 where f(y, θ) : R l+p R k is a map that is differentiable with respect to θ and rank(e[ f(y t, θ )/ θ ]) = p with k p. To deal with time series data, we consider the smoothed moment conditions introduced by Smith (211), f tn (θ) = 1 t 1 s=t n ( s K ) f(y t s, θ), (2.2) where K( ) is a kernel function and = bn with b (, 1) is the bandwidth parameter. Smoothing of the moment conditions induces a heteroskedasticity and autocorrelation consistent (HAC) covariance estimator of the long run variance matrix of {f(y t, θ)} n t=1. Let f t(θ) = f(y t, θ) and f n (θ) = n t=1 f tn(θ)/n, where f tn (θ) is defined in (2.2). Consider the profile empirical log-likelihood function based on the smoothed moment restrictions, { n L n (θ) = sup log(p t ) : p t, t=1 n p t = 1, t=1 n t=1 } p t f tn (θ) =. (2.3) Standard Lagrange multiplier arguments imply that the maximum is attained when 1 n p t = n{1 + λ f tn (θ)}, with f tn (θ) 1 + λ f tn (θ) =. The maximum empirical likelihood estimate (MELE) is then given by ˆθ el = argmax θ Θ L n (θ). Following Kitamura (26), the empirical log-likelihood function can also be derived by considering the dual problem (see e.g., Borwein and Lewis (1991)), L n (θ) = min λ R k t=1 n log(1 + λ f tn (θ)) n log n, (2.4) t=1 where log(x) = for x <. Here (2.4) has a natural connection with the generalized empirical likelihood (GEL), and it facilitates our theoretical derivation under the fixed-b asymptotics. To introduce the fixed-b approach, we define the empirical log-likelihood ratio function elr(θ) = 2 max λ R k n t=1 log(1 + λ f tn (θ)), (2.5) for θ Θ and = bn. Under the traditional small-b asymptotics, nb 2 +1/(nb) as n, and suitable weak dependence assumptions (see e.g., Smith (211)), it can be shown that elr(θ ) = n f n (θ ) ( b n t=1 f tn (θ )f tn (θ ) ) 1 fn (θ ) + o p (1) d κ2 1 κ 2 χ 2 k,
4 4 XIANYANG ZHANG AND XIAOFENG SHAO where κ 1 = + K(x)dx and κ 2 = + K2 (x)dx (assuming that κ 1, κ 2 < ). However, the χ 2 -based approximation can be poor, especially when the dependence is strong and the bandwidth parameter is large (see Section 3). To derive the fixed-b limiting distribution, we make an assumption that is standard in moment condition models. Assumption 1. nr t=1 f t(θ )/ n ΛW k (r) for r [, 1], where ΛΛ = Ω = + j= Γ j with Γ j = Ef t+j (θ )f t (θ ), and W k (r) is a k-dimensional vector of independent standard Brownian motions. Assumption 1 can be verified under suitable moment and weak dependence assumptions on f t (θ ) (see e.g., Phillips (1987)). For the kernel function, we assume the following. Assumption 2. The kernel K : R [ c, c ] for some < c <, is piecewise continuously differentiable. Fix b (, 1), where b = /n. Using summation by parts, the Continuous Mapping Theorem and Itô s formula, it is not hard to show that, for t = nr with r [, 1], nftn (θ ) = n t 1 s=t n ( s ) K f t s (θ ) ΛD k(r; b), (2.6) b where D k (r; b) = K((r s)/b)dw k(s). Let C k [, 1] = {(f 1, f 2,..., f k ) : f i C[, 1]}. For any g C k [, 1], take G el (g) = max λ R k log(1 + λ g(t))dt. We show in the Appendix that the functional G el ( ) is continuous under the sup norm. Therefore, by the Continuous Mapping Theorem, we can characterize the asymptotic behavior of elr(θ ). Theorem 1. Suppose Assumptions 1 2 hold. For n + and b fixed, elr(θ ) d U el,k (b; K) := 2 ( r s ) ) b max log (1 + λ K dw k (s) dr. (2.7) λ R k b The proof of Theorem 1 is given in the supplementary material. Theorem 1 shows that the fixed-b limiting distribution of elr(θ ) is nonstandard yet pivotal for a given bandwidth and kernel, and its critical values can be obtained via simulation or iid bootstrap (because the bootstrapped sample satisfies the Functional Central Limit Theorem). Let u el,k (b; K; 1 α) be the 1(1 α)% quantile of U el,k (b; K)/(1 b). Given b (, 1), a 1(1 α)% confidence region for the parameter θ is then given by { CI(1 α; b) = θ R p : elr(θ) } 1 b u el,k(b; K; 1 α). (2.8)
5 FIXED-b ASYMPTOTICS FOR BLOCKWISE EMPIRICAL LIKELIHOOD 5 When K(x) = I(x ), we have D k (r; b) = W k (r) and A := {λ R k : min r [,1] (1 + λ D k (r; b)) } = {λ R k : min r [,1] (1 + λ W k (r)) }. By Lemma 1 of Nordman, Bunzel, and Lahiri (213), we know that A is bounded with probability one, which implies that P (U el,k (b; K) = ) =. We conjecture that P (U el,k (b; K) = ) can be positive for particular K( ) and b (, 1). In our simulations, critical values are calculated based on the cases where U el,k (b; K) < (when b is close to zero, P (U el,k (b; K) = ) is rather small, as seen from our unreported simulation results). The nonstandard limiting distribution also provides some insights on how likely the origin is not contained in the convex hull of {f tn (θ )} n t=1 when the sample size n is large. Remark 1. To capture the dependence within the observations, one can employ the commonly used blocking technique first applied to the EL by Kitamura (1997). To illustrate, we consider the fully overlapping smoothed moment condition given by f tn (θ) = (1/m) t+m 1 j=t f(y j, θ) with t = 1, 2,..., n m + 1 and m = nb for b (, 1). Under suitable weak dependence assumptions, we have nftn (θ ) Λ{W k (r + b) W k (r)}/b for t = nr. Using similar arguments to those in Theorem 1, we can show that elr(θ ) d U el,k (b) := 2 b max λ R k b log(1 + λ {W k (r + b) W k (r)})dr. We generate the critical values of U el,k (b)/(1 b) (conditioning on U el,k (b) < ) for b from.1 to.3 with spacing.1, and further approximate the critical values by a cubic function of b following the practice of Kiefer and Vogelsang (25). The estimates of the coefficients of the corresponding cubic functions are given in Table 1. Similarly, we summarize the critical values of U el,k (b; K)/(1 b) (conditioning on U el,k (b; K) < ) with K(x) = (5π/8) 1/2 (1/x)J 1 (6πx/5) for b from.1 to.2 in Table 2, where J 1 ( ) denotes the Bessel function of the first kind. Remark 2. A natural question to ask is whether the fixed-b asymptotics is consistent with the traditional small-b asymptotics when b is close to zero. We provide an affirmative answer by showing that U el,k (b, K) converges to a scaled χ 2 k distribution as b. We assume K satisfies certain regularity conditions (see Assumption 2.2 in Smith (211)). Using the Taylor expansion and some standard arguments for EL, it is not hard to show that U el,k (b; K) = 1 b ( ) 1 D k (r, b) dr D k (r; b)d k (r; b) dr D k (r, b)dr + o p (1).
6 6 XIANYANG ZHANG AND XIAOFENG SHAO Table 1. Critical value function coefficients. a a 1 a 2 a 3 R 2 u el,1 (b;.9) u el,1 (b;.95) u el,1 (b;.99) u el,2 (b;.9) u el,2 (b;.95) u el,2 (b;.99) u el,3 (b;.9) u el,3 (b;.95) u el,3 (b;.99) The critical value u el,k (b; 1 α) is approximated by a cubic function a + a 1b + a 2b 2 + a 3b 3 of b. The estimated coefficients and multiple R 2 are reported. The Brownian motion is approximated by a normalized partial sum of 1, i.i.d. standard normal random variables and the number of Monte Carlo replication is 5,. Table 2. Critical value function coefficients. a a 1 a 2 a 3 R 2 u el,1 (b; K;.9) u el,1 (b; K;.95) u el,1 (b; K;.99) u el,2 (b; K;.9) u el,2 (b; K;.95) u el,2 (b; K;.99) u el,3 (b; K;.9) u el,3 (b; K;.95) u el,3 (b; K;.99) The critical value u el,k (b; K; 1 α) is approximated by a cubic function a + a 1 b + a 2 b 2 + a 3 b 3 of b. The estimated coefficients and multiple R 2 are reported. The Brownian motion is approximated by a normalized partial sum of 1, iid standard normal random variables and the number of Monte Carlo replication is 5,. Under Assumption 2, we derive that 1 b D k (r, b)dr = = ( r s K b (1 s)/b s/b ) drd W k(s) b K(t)dtdW k (s) d κ 1 W k (1). Define the semi-positive definite kernel K b (r, s)= K((t r)/b)k((t s)/b)dt/(bκ 2).
7 FIXED-b ASYMPTOTICS FOR BLOCKWISE EMPIRICAL LIKELIHOOD 7 Then V k (b) = (1/b) D k(r; b)d k (r; b) dr = κ 2 K b (r, s)dw k(r)dw k (s) = d κ + 2 j=1 λ j,bη j η j, where {η j} + j=1 is an independent sequence of N k(, I k ) random vectors and the λ j,b are the eigenvalues associated with Kb (r, s). Note that E{V k (b)} 1 = I k Kb κ (r, r)dr = I k K 2( t r ) dtdr I k. 2 κ 2 b b Let η j = (η j1,..., η jk ) and denote by V (l,m) k (b) the (l, m)th element of V k (b) with 1 l, m k. Since + j=1 λ2 j,b = {K b (r, s)}2 drds as b (see e.g., Sun (21)), we get { (l,m) V k (b)} 2 + E = κ 2 = + j=1 j =1 + j=1 λ2 j,b ( + j=1 λ j,b λ j,b λ j,beη jl η jm η j lη j m, l m; ) , l = m, j=1 λ2 j,b which implies that V k (b) p κ 2 I k. Therefore,we have U el,k (b, K) d (κ 2 1 /κ 2)χ 2 k as b. Compared to the χ 2 -approximation, the fixed-b limiting distribution that captures the choice of the kernel and the bandwidth is expected to provide better approximation to the finite sample distribution of the BEL ratio statistic at the true parameter when b is relatively large Generalized empirical likelihood We extend the fixed-b approach to the Generalized empirical likelihood (GEL) estimation framework (Newey and Smith (24)). To describe GEL, we let ρ be a concave function defined on an open set I that contains the origin. Set ρ(x) = for x / I, and let ρ j (x) = j ρ(x)/ x j and ρ j = ρ j () for j =, 1, 2. We normalize ρ so that ρ 1 = ρ 2 = 1. Consider the set Π n (θ) = {λ : λ f tn (θ) I, t = 1, 2,..., n}. The GEL estimator is the solution to a saddle point problem, ˆθ gel = argmin θ Θ sup λ R k ˆP (θ, λ) = argmin θ Θ sup ˆP (θ, λ), λ Π n (θ) where ˆP (θ, λ) = 1 n t=1 {ρ(λ f tn (θ)) ρ }. The GEL ratio function is given by gelr(θ) = 2 sup λ R k ˆP (θ, λ). (2.9) The GEL estimator includes a number of special cases that have been well studied in the statistics and econometrics literature. The EL, exponential tilting (ET), and continuous updating (CUE) are special cases of the GEL. Thus
8 8 XIANYANG ZHANG AND XIAOFENG SHAO ρ(x) = log(1 x) and I = (, 1) for EL, ρ(x) = e x and I = R for ET, and ρ(x) = (1 + x) 2 /2 and I = R for CUE. More generally, members of the Cressie- Read power divergence family of discrepancies discussed by Imbens, Spady, and Johnson (1998) are included in the GEL class with ρ(x) = (1+γx) (γ+1)γ /(γ+1) (see Newey and Smith (24)). {ρ(λ g(t)) ρ }dt for g C k [, 1]. If ρ( ) is Let G gel (f) = max λ R k strictly concave and twice continuously differentiable, under suitable assumptions it can be shown that G gel ( ) is a continuous functional under the sup norm. Since the argument follows from that presented in the appendix with a minor modification, we skip the details (see Remark 1.1 in the supplementary material). Therefore, we have gelr(θ ) d U ρ,k (b; K) := 2 b max λ R k { ) } ρ (λ K((r s)/b)dw k (s) ρ dr. The GEL-based confidence region for the parameter θ is { CI(1 α; b) = θ R p : gelr(θ) } 1 b u ρ,k(b; K; 1 α), (2.1) where u ρ,k (b; K; 1 α) is the 1(1 α)% quantile of U ρ,k (b; K)/(1 b), which can again be obtained via simulation or iid bootstrap. 3. Numerical Studies We conducted two sets of simulation studies to compare and contrast the finite sample performance of the inference procedure based on the fixed-b approximation and the BEL of Kitamura (1997) and Smith (211). The simulation results presented below are based on the simulation runs where the origin is contained in the convex hull of {f tn (θ )} Mean and quantiles Consider the time series models AR(1), Y t = ρy t 1 + ϵ t with ρ =.5,.2,.5,.8, and AR(2), Y t = (5/6)Y t 1 (1/6)Y t 2 +ϵ t. The latter was used in Chen and Wong (29) where the focus was to compare the finite sample coverages of the quantile delivered by BEL. In both models, {ϵ t } is a sequence of iid standard normal random variables. We focus on the inference for the mean, the median and the 5% quantile. For the mean, f(y t, θ) = y t θ. For the q-th quantile, we consider the moment condition f q (y t, θ) = (θ y t )/h K(x)dx q, where K( ) is an r-th order window that satisfies 1, j = ; u j K(u)du =, 1 j r 1; κ, j = r,
9 FIXED-b ASYMPTOTICS FOR BLOCKWISE EMPIRICAL LIKELIHOOD 9 for some integer r 2, and h is a bandwidth such that h as n +. When h =, we have f q (y t, θ) = I(y t θ) q. To accommodate dependence, we consider the BEL with fully overlapping moment conditions f tn (θ) = (1/m) t+m 1 j=t f(y t, θ) for t = 1, 2,..., n m + 1 and m = nb with b (, 1). For comparisons, we consider smoothed EL with the kernel K(x) = (5π/8) 1/2 (1/x)J 1 (6πx/5), where J 1 ( ) is the Bessel function of the first kind. The HAC covariance estimator induced by using K( ) is essentially the same as the nonparametric long run variance estimator with the Quadratic spectral kernel (see Example 2.3 of Smith (211)). The sample sizes considered were n = 1 and 4, and b was chosen from.2 to.2. To draw inference for the quantiles, we employed the second order Epanechnikov window with bandwidth h = cn 1/4 for c =, 1, following Chen and Wong (29). The coverage probabilities and corresponding interval widths for the mean and quantiles delivered by the fixed-b approximation and the χ 2 -based approximation are depicted in Figures 1 4. For the mean, undercoverage occurs for both the fixed-b calibration and the χ 2 -based approximation when the dependence is positive, and becomes more severe as the dependence strengthens. Inference based on the fixed-b calibration provided uniformly better coverage probabilities in all cases, and was quite robust to the choice of b. The improvement was significant, especially for large bandwidth. On the other hand, the fixed-b based interval was slightly wider than the χ 2 -based interval. For negative dependence with ρ =.5, the fixed-b calibration tended to provide overcoverage, but the improvement over the χ 2 -based approximation could be seen for relatively large b. These findings are consistent with the intuition that the larger b is, the more accurate the fixed-b based approximation is relative to the χ 2 -based approximation used by Kitamura (1997) and Smith (211). The results for the median and 5% quantile were qualitatively similar to those in the mean case. The choice of h = 1 tended to provide slightly shorter interval widths as compared to the unsmoothed counterpart, h = in some cases (see Chen and Wong (29)). A comparison of Figure 1 with Figure 3 (Figure 2 with Figure 4) has the coverage probabilities for the EL based on the kernel K(x) generally closer to the nominal level than the BEL counterpart with the corresponding interval widths wider. This phenomenon is consistent with the finding that QS kernel provides better coverage, but wider interval widths, compared to the Bartlett kernel in Kiefer and Vogelsang (25) under the GMM framework. Our unreported simulation results also demonstrate the usefulness of fixed-b calibration under the GEL estimation framework. The results for ET are available upon request.
10 1 XIANYANG ZHANG AND XIAOFENG SHAO Figure 1. Coverage probabilities for the mean delivered by the BEL based on the fixed-b approximation and the χ 2 -based approximation. The nominal level is 95% and the number of Monte Carlo replications is 1,.
11 FIXED-b ASYMPTOTICS FOR BLOCKWISE EMPIRICAL LIKELIHOOD 11 Figure 2. Coverage probabilities for the median and 5% quantile delivered by the BEL based on the fixed-b approximation and the χ 2 -based approximation. The nominal level is 95% and the number of Monte Carlo replications is 1,.
12 12 XIANYANG ZHANG AND XIAOFENG SHAO Figure 3. Coverage probabilities for the mean delivered by the smoothed EL based on the fixed-b approximation and the χ 2 -based approximation. The corresponding kernel is K(x) = (5π/8) 1/2 (1/x)J 1 (6πx/5). The nominal level is 95% and the number of Monte Carlo replications is 1,.
13 FIXED-b ASYMPTOTICS FOR BLOCKWISE EMPIRICAL LIKELIHOOD 13 Figure 4. Coverage probabilities for the median and 5% quantile delivered by the smoothed EL based on the fixed-b approximation and the χ 2 -based approximation. The corresponding kernel is K(x) = (5π/8) 1/2 (1/x)J 1 (6πx/5). The nominal level is 95% and the number of Monte Carlo replications is 1,.
14 14 XIANYANG ZHANG AND XIAOFENG SHAO Figure 5. Coverage probabilities delivered by the BEL (left panels) and the smoothed EL (right panels) based on the fixed-b approximation and the χ 2 -based approximation. The corresponding kernel for the smoothed EL is K(x) = (5π/8) 1/2 (1/x)J 1 (6πx/5). The nominal level is 95% and the number of Monte Carlo replications is 1, Time series regression We consider the stylized linear regression model with an intercept and a regressor x t : y t = β 1 + β 2 x t + u t for 1 t n, where {x t } and {u t } are generated independently from an AR(1) model with common coefficient ρ. We set the true parameter β = (β 1, β 2 ) = (, ) and chose ρ {.2,.5,.8}. We are interested in constructing confidence contour for β. Consider the moment conditions f t (β) = (u t (β), x t u t (β), x t 1 u t (β), x t 2 u t (β)) with u t (β) = y t β 1 β 2 x t and 3 t n. We report the coverage probabilities for the BEL and the
15 FIXED-b ASYMPTOTICS FOR BLOCKWISE EMPIRICAL LIKELIHOOD 15 smoothed EL with kernel K(x) based on the fixed-b approximation and the χ 2 - based approximation in Figure 5. As the dependence strengthens, the fixed-b and χ 2 -based approximations deteriorate. The coverage probabilities obtained from the fixed-b calibration are consistently closer to the nominal level, and the improvement is significant for large bandwidths. In contrast, the coverage probabilities based on the χ 2 approximation are severely downward biased for relatively large b. To sum up, the fixed-b approximation provides a uniformly better approximation to the sampling distribution of the EL ratio statistic for a wide range of b, and it tends to deliver more accurate coverage probability in confidence interval construction and size in testing. From a practical viewpoint, the choice of the bandwidth parameter has a great impact on the finite sample performance of the EL ratio statistic; it is of interest to consider the optimal bandwidth under the fixed-b paradigm. Acknowledgements We are grateful to two referees for their helpful comments that led to substantial improvements. Shao s research is supported in part by National Science Foundation grant DMS References Billingsley, P. (1999). Convergence of Probability Measures. Second edition. Wiley, New York. Borwein, J. M. and Lewis, A. S. (1991). Duality relationships for entropy-type minimization problems. SIAM J. Control Optim. 29, Chen, S. X. and Van Keilegom, I. (29). A review on empirical likelihood methods for regression. TEST 18, Chen, S. X. and Wong, C. (29). Smoothed block empirical likelihood for quantiles of weakly dependent processes. Statist. Sinica 19, Cressie, N. and Read, T. (1984). Multinomial goodness-of-fit tests. J. Roy. Statist. Soc. Ser. B 46, Imbens, G. W., Spady, R. H. and Johnson, P. (1998). Information theoretic approaches to inference in moment condition models. Econometrica 66, Jansson, M. (24). On the error of rejection probability in simple autocorrelation robust tests. Econometrica 72, Kiefer, N. M. and Vogelsang, T. J. (25). A new asymptotic theory for heteroskedasticityautocorrelation robust tests. Econometric Theory 21, Kitamura, Y. (1997). Empirical likelihood methods with weakly dependent processes. Ann. Statist. 25, Kitamura, Y. (26). Empirical likelihood methods in econometrics: Cowles Foundation Discussion Paper theory and practice. Newey, W. K. and Smith, R. (24). Higher order properties of GMM and generalized empirical likelihood estimators. Econometrica 72,
16 16 XIANYANG ZHANG AND XIAOFENG SHAO Nordman, D. J. (29). Tapered empirical likelihood for time series data in time and frequency domains. Biometrika 96, Nordman, D. J., Bunzel, H. and Lahiri, S. N. (213). A non-standard empirical likelihood for time series. Ann. Statist. To appear. Owen, A. (1988). Empirical likelihood ratio confidence intervals for a single functional. Biometrika 75, Owen, A. (199). Empirical likelihood confidence regions. Ann. Statist. 18, Owen, A. (21). Empirical Likelihood. Chapman and Hall, New York. Phillips, P. C. B. (1987). Time series regression with unit roots. Econometrica 55, Qin, J. and Lawless, J. (1994). Empirical likelihood and general estimating equations. Ann. Statist. 22, Shao, X. and Politis, D. N. (213). Fixed-b subsampling and block bootstrap: improved confidence sets based on p-value calibration. J. Roy. Statist. Soc. Ser. B 75, Smith, R. (211). GEL criteria for moment condition models. Econometric Theory 27, Sun, Y. (21). Let s fix it: fixed-b asymptotics versus small-b asymptotics in heteroscedasticity and autocorrelation robust Inference. working paper. Sun, Y., Phillips, P. C. B. and Jin, S. (28). Optimal bandwidth selection in heteroscedasicityautocorrlation robust testing. Econometrica 76, Zhang, X. and Shao, X. (213). Fixed-smoothing asymptotics for time series. Ann. Statist. 41, Department of Statistics, University of Missouri-Columbia, Columbia, MO 65211, USA. zhangxiany@missouri.edu Department of Statistics, University of Illinois at Urbana-Champaign, 725 S. Wright St., Champaign, IL 6182, USA. xshao@illinois.edu (Received November 212; accepted July 213)
17 Statistica Sinica: Supplement Fixed-b Asymptotics for Blockwise Empirical Likelihood Xianyang Zhang and Xiaofeng Shao University of Missouri-Columbia and University of Illinois at Urbana-Champaign Supplementary Material The following supplementary material contains the proof of Theorem 2.1. S1 Technical appendix Define the set of functions Q = {g = (g 1, g 2,..., g k ) C k [, 1] : g is are linearly independent}, and let G el (g) = max λ R k log(1 + λ g(t))dt be a nonlinear functional from C k [, 1] to the real line R, where log(x) = for x <. We shall prove in the following that G el (g) is a continuous map for functions in Q under the sup norm (Seijo and Sen (211)). For any g C k [, 1], we define H g = {λ R k : min t [,1] (1 + λ g(t)) } and L g (λ) = log(1 + λ g(t))dt. It is straightforward to show that L g (λ) is strictly convex for g Q on the set H g. We also note that H g is a closed convex set, which contains a neighborhood of the origin. Let λ g = argmax λ R k log(1 + λ g(t))dt be the maximizer of L g (λ). We first show that G el (g) < if and only if H g is bounded. If G el (g) =, then λ g cannot be finite, which implies that H g is unbounded. On the other hand, suppose H g is unbounded. Note that H g = t [,1] {λ R k : λ g(t) 1} which is the intersection of a set of closed half-spaces. The recession cone of H g is then given by + H g = t [,1] {λ R k : λ g(t) } (see Section 8 of Rockafellar (197)). By Theorem 8.4 of Rockafellar (197), there exists a nonzero vector λ + H g, and the set {t [, 1] : λ g(t) > } has positive Lebesgue measure because of the linearly independence of g. We have G el (g) L g (a λ) for any a >, where L g (a λ) as a. Thus we get G el (g) =. Next, we consider the case G el (g) =. Following the discussion above, there exists δ such that the set B := {t [, 1] : λ g(t) > δ} has Lebesgue measure Λ(B) >. For any A >, we choose ɛ (, 1) and large enough a > so that Λ(B) log(1 + a δ ɛ ) + log(1 ɛ ) > A.
18 S2 Xianyang Zhang and Xiaofeng Shao For any f Q with f g := sup t [,1] f(t) g(t) ɛ /( λ a), we have log(1 + a λ f(t))dt = log(1 + a λ (f(t) g(t)) + a λ g(t))dt B + log(1 + a λ (f(t) g(t)) + a λ g(t))dt B c Λ(B) log(1 + a δ ɛ ) + log(1 ɛ ) > A. In what follows, we turn to the case G el (g) <, i.e., H g is bounded as shown before. Case 1: we first consider the case that λ g H g = {λ R k : min t [,1] (1 + λ g(t)) > }. Since H g is open, we can pick a positive number τ so that B(λ g ; τ) := {λ R k : λ λ g τ} H g. Then we have min λ B(λg;τ) min t [,1] (1 + λ g(t)) > c >. Furthermore, there exists a sufficiently small δ such that for any f Q with f g δ, we have min t [,1] (1 + λ f(t)) > c > for any λ B(λ g ; τ), i.e., B(λg ; τ) H f. Notice that the constant c only depends on g, δ and c. Given any ɛ >, we shall first show that sup λ B(λg;τ) L f (λ) L g (λ) < ɛ for any f Q with f g < δ(ɛ), where < δ(ɛ) < δ. Because G el (g) <, we have log(1 + λ g(t))dt < for any λ B(λ g ; τ). Simple algebra yields that log(1 + λ f(t))dt log(1 + λ g(t))dt { max log(1 + M δ(ɛ)/c ), log(1 + M δ(ɛ)/c) }, (S1.1) where M = λ g + τ. The RHS of (1) can be made arbitrarily small for sufficiently small δ(ɛ). Therefore we get sup λ B(λg;τ) L f (λ) L g (λ) < ɛ for small enough δ(ɛ), which implies that G el (g) sup λ B(λg;τ) log(1 + λ f(t))dt < ɛ. Next, we show that there exists a local maxima of L f (λ) in B(λ g ; τ). Suppose ɛ is sufficiently small and choose < ξ < τ such that L g (λ g ) > max λ B(λg;τ) B c (λ g;ξ) L g (λ) + 2ɛ, where B(λ g ; ξ) = {λ R k : λ λ g < ξ}. Thus we get max λ B(λ g;τ) B c (λ g;ξ) L f (λ) max λ B(λ g;τ) B c (λ g;ξ) L g (λ) + ɛ < L g (λ g ) ɛ L f (λ g ) max L f (λ). λ B(λ g;ξ) Because f Q, L f (λ) is strictly convex. Hence, the local maxima is also the global maxima, which implies that G el (g) G el (f) < ɛ. Case 2: We now consider the case min t [,1] (1 + λ gg(t)) =. For any < δ < δ < 1, let H g (δ ) = {(1 δ )λ : λ H g } and H f (δ ) = {(1 δ )λ : λ H f }. There exists a small enough δ > such that for any f Q with f g < δ, H f (δ ) H g (δ ) H f H g. By the continuity of L g (λ), we know for any ɛ >, there exists a δ > such that when λ λ g δ λ g, L g (λ g ) < L g (λ) + ɛ/4. By the construction
19 Fixed-b Asymptotics for Blockwise Empirical Likelihood S3 of H g (δ ), we have L g (λ g ) < L g ((1 δ )λ g ) + ɛ/4 sup L g (λ) + ɛ/4. λ H g(δ ) Using similar arguments in the first case and the boundness of H g, we can show that sup L g (λ) sup L f (λ) < ɛ/8, λ H g(δ ) λ H g(δ ) for sufficiently small δ. Furthermore, when λ f H g (δ ), we have L f (λ f ) = sup λ Hg(δ ) L f (λ). When λ f / H g (δ ), by the convexity of L f (λ), we get L f ((1 δ )λ f ) (1 δ )L f (λ f ), which implies that L f (λ f ) L f ((1 δ )λ f ) 1 δ sup λ H g(δ ) L f (λ) 1 δ sup λ H g(δ ) L g (λ) + ɛ/8 1 δ sup L g (λ) + ɛ/4 λ H g(δ ) < sup L f (λ) + ɛ/2 λ H g(δ ) for small enough δ (e.g., δ ɛ < min(1/3, 24G el (g))). Thus we have G el (f) G el (g) L f (λ f ) sup L f (λ) λ H g(δ ) + L g(λ g ) sup L g (λ) sup L f (λ) < ɛ. λ H g(δ ) λ H g(δ ) sup λ H g(δ ) L g (λ) Combining the above arguments, we show that the map G el is continuous under the sup norm. Next, we consider the limiting process D k (r; b) = K((r s)/b)dw k(s) with b (, 1) being fixed in the asymptotics. Because the components of D k (r; b) are mutually independent, we have P (α D k (r; b) = for some α R k ) = which implies that P (D k (r; b) Q) = 1. Under the assumptions in Theorem 2.1, the set {λ : min r [,1] (1 + λ D k (r; b)) } is compact and convex almost surely (note the convexity and closeness of the set follow directly from its definition). Using summation by parts, we get n nftn (θ ) = = 1 ( t n b n K ) n t 1 s=t n k=1 ( ) s n K f t s (θ ) = n 1 ( t s f k (θ ) + 1 b n s=1 { K n ( t s K s=1 ) K ) f s (θ ) ( )} s t s 1 f k (θ ). k=1
20 S4 Xianyang Zhang and Xiaofeng Shao By the continuous mapping theorem and Itô s formula, we obtain { ( ) 1 r 1 nftn (θ ) d Λ b K W k (1) + 1b ( ) } r s b 2 K W k (s)ds = d ΛD k (r; b)/b, b for t = nr with r [, 1]. Finally, by the continuous mapping theorem, we get elr(θ ) = 2 b max λ R k n log(1 + λ nbλ 1 f tn (θ ))/n, t=1 d U el,k (b; K) := 2 b max λ R k ( log 1 + λ ) D k (r; b) dr. λ = Λ λ/( nb), (S1.2) Remark S1.1. For ET and CUE, we have I = R. Given any g Q with G gel (g) <, we have H g = {λ R k : λ g(t) I, for all t [, 1]} = R k and λ g <. Therefore, λ g is an interior point of H g and the arguments in Case 1 can be applied to show the continuity of G gel ( ) at g. References Rockafellar, T. R. (197). Convex Analysis. Princeton Univ. Press. Seijo, E. and Sen, B. (211). A continuous mapping theorem for the smallest argmax functional. Electron. J. Stat. 5,
SMOOTHED BLOCK EMPIRICAL LIKELIHOOD FOR QUANTILES OF WEAKLY DEPENDENT PROCESSES
Statistica Sinica 19 (2009), 71-81 SMOOTHED BLOCK EMPIRICAL LIKELIHOOD FOR QUANTILES OF WEAKLY DEPENDENT PROCESSES Song Xi Chen 1,2 and Chiu Min Wong 3 1 Iowa State University, 2 Peking University and
More informationHigh Dimensional Empirical Likelihood for Generalized Estimating Equations with Dependent Data
High Dimensional Empirical Likelihood for Generalized Estimating Equations with Dependent Data Song Xi CHEN Guanghua School of Management and Center for Statistical Science, Peking University Department
More informationA New Approach to Robust Inference in Cointegration
A New Approach to Robust Inference in Cointegration Sainan Jin Guanghua School of Management, Peking University Peter C. B. Phillips Cowles Foundation, Yale University, University of Auckland & University
More informationAdjusted Empirical Likelihood for Long-memory Time Series Models
Adjusted Empirical Likelihood for Long-memory Time Series Models arxiv:1604.06170v1 [stat.me] 21 Apr 2016 Ramadha D. Piyadi Gamage, Wei Ning and Arjun K. Gupta Department of Mathematics and Statistics
More informationTesting in GMM Models Without Truncation
Testing in GMM Models Without Truncation TimothyJ.Vogelsang Departments of Economics and Statistical Science, Cornell University First Version August, 000; This Version June, 001 Abstract This paper proposes
More informationWhat s New in Econometrics. Lecture 15
What s New in Econometrics Lecture 15 Generalized Method of Moments and Empirical Likelihood Guido Imbens NBER Summer Institute, 2007 Outline 1. Introduction 2. Generalized Method of Moments Estimation
More informationSpring 2017 Econ 574 Roger Koenker. Lecture 14 GEE-GMM
University of Illinois Department of Economics Spring 2017 Econ 574 Roger Koenker Lecture 14 GEE-GMM Throughout the course we have emphasized methods of estimation and inference based on the principle
More informationLECTURE ON HAC COVARIANCE MATRIX ESTIMATION AND THE KVB APPROACH
LECURE ON HAC COVARIANCE MARIX ESIMAION AND HE KVB APPROACH CHUNG-MING KUAN Institute of Economics Academia Sinica October 20, 2006 ckuan@econ.sinica.edu.tw www.sinica.edu.tw/ ckuan Outline C.-M. Kuan,
More informationGeneralized Neyman Pearson optimality of empirical likelihood for testing parameter hypotheses
Ann Inst Stat Math (2009) 61:773 787 DOI 10.1007/s10463-008-0172-6 Generalized Neyman Pearson optimality of empirical likelihood for testing parameter hypotheses Taisuke Otsu Received: 1 June 2007 / Revised:
More informationThe Bootstrap: Theory and Applications. Biing-Shen Kuo National Chengchi University
The Bootstrap: Theory and Applications Biing-Shen Kuo National Chengchi University Motivation: Poor Asymptotic Approximation Most of statistical inference relies on asymptotic theory. Motivation: Poor
More informationResearch Article Optimal Portfolio Estimation for Dependent Financial Returns with Generalized Empirical Likelihood
Advances in Decision Sciences Volume 2012, Article ID 973173, 8 pages doi:10.1155/2012/973173 Research Article Optimal Portfolio Estimation for Dependent Financial Returns with Generalized Empirical Likelihood
More informationBootstrapping high dimensional vector: interplay between dependence and dimensionality
Bootstrapping high dimensional vector: interplay between dependence and dimensionality Xianyang Zhang Joint work with Guang Cheng University of Missouri-Columbia LDHD: Transition Workshop, 2014 Xianyang
More informationSerial Correlation Robust LM Type Tests for a Shift in Trend
Serial Correlation Robust LM Type Tests for a Shift in Trend Jingjing Yang Department of Economics, The College of Wooster Timothy J. Vogelsang Department of Economics, Michigan State University March
More informationSelf-normalization for Time Series: A Review of Recent Developments 1
Self-normalization for Time Series: A Review of Recent Developments 1 February 24, 2015 Xiaofeng Shao Abstract: This article reviews some recent developments on the inference of time series data using
More informationThe properties of L p -GMM estimators
The properties of L p -GMM estimators Robert de Jong and Chirok Han Michigan State University February 2000 Abstract This paper considers Generalized Method of Moment-type estimators for which a criterion
More information11. Bootstrap Methods
11. Bootstrap Methods c A. Colin Cameron & Pravin K. Trivedi 2006 These transparencies were prepared in 20043. They can be used as an adjunct to Chapter 11 of our subsequent book Microeconometrics: Methods
More informationA Fixed-b Perspective on the Phillips-Perron Unit Root Tests
A Fixed-b Perspective on the Phillips-Perron Unit Root Tests Timothy J. Vogelsang Department of Economics Michigan State University Martin Wagner Department of Economics and Finance Institute for Advanced
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 informationICSA Applied Statistics Symposium 1. Balanced adjusted empirical likelihood
ICSA Applied Statistics Symposium 1 Balanced adjusted empirical likelihood Art B. Owen Stanford University Sarah Emerson Oregon State University ICSA Applied Statistics Symposium 2 Empirical likelihood
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 informationHAR Inference: Recommendations for Practice
HAR Inference: Recommendations for Practice Eben Lazarus, Harvard Daniel Lewis, Harvard Mark Watson, Princeton & NBER James H. Stock, Harvard & NBER JBES Invited Session Sunday, Jan. 7, 2018, 1-3 pm Marriott
More informationSupplement to Quantile-Based Nonparametric Inference for First-Price Auctions
Supplement to Quantile-Based Nonparametric Inference for First-Price Auctions Vadim Marmer University of British Columbia Artyom Shneyerov CIRANO, CIREQ, and Concordia University August 30, 2010 Abstract
More informationGoodness-of-Fit Tests for Time Series Models: A Score-Marked Empirical Process Approach
Goodness-of-Fit Tests for Time Series Models: A Score-Marked Empirical Process Approach By Shiqing Ling Department of Mathematics Hong Kong University of Science and Technology Let {y t : t = 0, ±1, ±2,
More informationExponential Tilting with Weak Instruments: Estimation and Testing
Exponential Tilting with Weak Instruments: Estimation and Testing Mehmet Caner North Carolina State University January 2008 Abstract This article analyzes exponential tilting estimator with weak instruments
More informationOptimal Bandwidth Selection in Heteroskedasticity-Autocorrelation Robust Testing
Optimal Bandwidth Selection in Heteroskedasticity-Autocorrelation Robust Testing Yixiao Sun Department of Economics University of California, San Diego Peter C. B. Phillips Cowles Foundation, Yale University,
More informationAsymptotic inference for a nonstationary double ar(1) model
Asymptotic inference for a nonstationary double ar() model By SHIQING LING and DONG LI Department of Mathematics, Hong Kong University of Science and Technology, Hong Kong maling@ust.hk malidong@ust.hk
More informationSpatially Smoothed Kernel Density Estimation via Generalized Empirical Likelihood
Spatially Smoothed Kernel Density Estimation via Generalized Empirical Likelihood Kuangyu Wen & Ximing Wu Texas A&M University Info-Metrics Institute Conference: Recent Innovations in Info-Metrics October
More informationHeteroskedasticity and Autocorrelation Consistent Standard Errors
NBER Summer Institute Minicourse What s New in Econometrics: ime Series Lecture 9 July 6, 008 Heteroskedasticity and Autocorrelation Consistent Standard Errors Lecture 9, July, 008 Outline. What are HAC
More informationUniversity of California, Berkeley
University of California, Berkeley U.C. Berkeley Division of Biostatistics Working Paper Series Year 24 Paper 153 A Note on Empirical Likelihood Inference of Residual Life Regression Ying Qing Chen Yichuan
More informationBlock Bootstrap HAC Robust Tests: The Sophistication of the Naive Bootstrap
Block Bootstrap HAC Robust ests: he Sophistication of the Naive Bootstrap Sílvia Gonçalves Département de sciences économiques, CIREQ and CIRANO, Université de ontréal and imothy J. Vogelsang Department
More informationEstimation of the Conditional Variance in Paired Experiments
Estimation of the Conditional Variance in Paired Experiments Alberto Abadie & Guido W. Imbens Harvard University and BER June 008 Abstract In paired randomized experiments units are grouped in pairs, often
More informationComment on HAC Corrections for Strongly Autocorrelated Time Series by Ulrich K. Müller
Comment on HAC Corrections for Strongly Autocorrelated ime Series by Ulrich K. Müller Yixiao Sun Department of Economics, UC San Diego May 2, 24 On the Nearly-optimal est Müller applies the theory of optimal
More informationSize and Shape of Confidence Regions from Extended Empirical Likelihood Tests
Biometrika (2014),,, pp. 1 13 C 2014 Biometrika Trust Printed in Great Britain Size and Shape of Confidence Regions from Extended Empirical Likelihood Tests BY M. ZHOU Department of Statistics, University
More informationA Resampling Method on Pivotal Estimating Functions
A Resampling Method on Pivotal Estimating Functions Kun Nie Biostat 277,Winter 2004 March 17, 2004 Outline Introduction A General Resampling Method Examples - Quantile Regression -Rank Regression -Simulation
More informationQuantile methods. Class Notes Manuel Arellano December 1, Let F (r) =Pr(Y r). Forτ (0, 1), theτth population quantile of Y is defined to be
Quantile methods Class Notes Manuel Arellano December 1, 2009 1 Unconditional quantiles Let F (r) =Pr(Y r). Forτ (0, 1), theτth population quantile of Y is defined to be Q τ (Y ) q τ F 1 (τ) =inf{r : F
More informationAFT Models and Empirical Likelihood
AFT Models and Empirical Likelihood Mai Zhou Department of Statistics, University of Kentucky Collaborators: Gang Li (UCLA); A. Bathke; M. Kim (Kentucky) Accelerated Failure Time (AFT) models: Y = log(t
More informationA strong consistency proof for heteroscedasticity and autocorrelation consistent covariance matrix estimators
A strong consistency proof for heteroscedasticity and autocorrelation consistent covariance matrix estimators Robert M. de Jong Department of Economics Michigan State University 215 Marshall Hall East
More informationMore Powerful Tests for Homogeneity of Multivariate Normal Mean Vectors under an Order Restriction
Sankhyā : The Indian Journal of Statistics 2007, Volume 69, Part 4, pp. 700-716 c 2007, Indian Statistical Institute More Powerful Tests for Homogeneity of Multivariate Normal Mean Vectors under an Order
More informationOn the Long-Run Variance Ratio Test for a Unit Root
On the Long-Run Variance Ratio Test for a Unit Root Ye Cai and Mototsugu Shintani Vanderbilt University May 2004 Abstract This paper investigates the effects of consistent and inconsistent long-run variance
More informationCAE Working Paper # Fixed-b Asymptotic Approximation of the Sampling Behavior of Nonparametric Spectral Density Estimators
CAE Working Paper #06-04 Fixed-b Asymptotic Approximation of the Sampling Behavior of Nonparametric Spectral Density Estimators by Nigar Hashimzade and Timothy Vogelsang January 2006. Fixed-b Asymptotic
More informationA New Approach to the Asymptotics of Heteroskedasticity-Autocorrelation Robust Testing
A New Approach to the Asymptotics of Heteroskedasticity-Autocorrelation Robust Testing Nicholas M. Kiefer Timothy J. Vogelsang August, 2 Abstract Asymptotic theory for heteroskedasticity autocorrelation
More informationTime Series and Forecasting Lecture 4 NonLinear Time Series
Time Series and Forecasting Lecture 4 NonLinear Time Series Bruce E. Hansen Summer School in Economics and Econometrics University of Crete July 23-27, 2012 Bruce Hansen (University of Wisconsin) Foundations
More informationCONSTRUCTING NONPARAMETRIC LIKELIHOOD CONFIDENCE REGIONS WITH HIGH ORDER PRECISIONS
Statistica Sinica 21 (2011), 1767-1783 doi:http://dx.doi.org/10.5705/ss.2009.117 CONSTRUCTING NONPARAMETRIC LIKELIHOOD CONFIDENCE REGIONS WITH HIGH ORDER PRECISIONS Xiao Li 1, Jiahua Chen 2, Yaohua Wu
More informationExtending the two-sample empirical likelihood method
Extending the two-sample empirical likelihood method Jānis Valeinis University of Latvia Edmunds Cers University of Latvia August 28, 2011 Abstract In this paper we establish the empirical likelihood method
More informationOn the coverage bound problem of empirical likelihood methods for time series
J. R. Statist. Soc. B (06) On the coverage ound prolem of empirical likelihood methods for time series Xianyang Zhang University of Missouri Columia, USA and Xiaofeng Shao University of Illinois at Urana-Champaign,
More informationCan we do statistical inference in a non-asymptotic way? 1
Can we do statistical inference in a non-asymptotic way? 1 Guang Cheng 2 Statistics@Purdue www.science.purdue.edu/bigdata/ ONR Review Meeting@Duke Oct 11, 2017 1 Acknowledge NSF, ONR and Simons Foundation.
More informationEconomics Division University of Southampton Southampton SO17 1BJ, UK. Title Overlapping Sub-sampling and invariance to initial conditions
Economics Division University of Southampton Southampton SO17 1BJ, UK Discussion Papers in Economics and Econometrics Title Overlapping Sub-sampling and invariance to initial conditions By Maria Kyriacou
More informationA New Asymptotic Theory for Heteroskedasticity-Autocorrelation Robust Tests
A New Asymptotic Theory for Heteroskedasticity-Autocorrelation Robust Tests Nicholas M. Kiefer and Timothy J. Vogelsang Departments of Economics and Statistical Science, Cornell University April 2002,
More informationThe Restricted Likelihood Ratio Test at the Boundary in Autoregressive Series
The Restricted Likelihood Ratio Test at the Boundary in Autoregressive Series Willa W. Chen Rohit S. Deo July 6, 009 Abstract. The restricted likelihood ratio test, RLRT, for the autoregressive coefficient
More informationWorking Paper No Maximum score type estimators
Warsaw School of Economics Institute of Econometrics Department of Applied Econometrics Department of Applied Econometrics Working Papers Warsaw School of Economics Al. iepodleglosci 64 02-554 Warszawa,
More informationLasso Maximum Likelihood Estimation of Parametric Models with Singular Information Matrices
Article Lasso Maximum Likelihood Estimation of Parametric Models with Singular Information Matrices Fei Jin 1,2 and Lung-fei Lee 3, * 1 School of Economics, Shanghai University of Finance and Economics,
More informationc 2010 by Zhewen Fan. All rights reserved.
c 2010 by Zhewen Fan. All rights reserved. STATISTICAL ISSUES AND DEVELOPMENTS IN TIME SERIES ANALYSIS AND EDUCATIONAL MEASUREMENT BY ZHEWEN FAN DISSERTATION Submitted in partial fulfillment of the requirements
More informationModification and Improvement of Empirical Likelihood for Missing Response Problem
UW Biostatistics Working Paper Series 12-30-2010 Modification and Improvement of Empirical Likelihood for Missing Response Problem Kwun Chuen Gary Chan University of Washington - Seattle Campus, kcgchan@u.washington.edu
More informationTesting for a Change in Mean under Fractional Integration
J. Time Ser. Econom. 206; aop Fabrizio Iacone*, Stephen J. Leybourne and A.M. Robert Taylor Testing for a Change in Mean under Fractional Integration DOI 0.55/jtse-205-0006 Abstract: We consider testing
More informationEmpirical Likelihood Methods for Sample Survey Data: An Overview
AUSTRIAN JOURNAL OF STATISTICS Volume 35 (2006), Number 2&3, 191 196 Empirical Likelihood Methods for Sample Survey Data: An Overview J. N. K. Rao Carleton University, Ottawa, Canada Abstract: The use
More informationEmpirical Likelihood Inference for Two-Sample Problems
Empirical Likelihood Inference for Two-Sample Problems by Ying Yan A thesis presented to the University of Waterloo in fulfillment of the thesis requirement for the degree of Master of Mathematics in Statistics
More informationOptimal Jackknife for Unit Root Models
Optimal Jackknife for Unit Root Models Ye Chen and Jun Yu Singapore Management University October 19, 2014 Abstract A new jackknife method is introduced to remove the first order bias in the discrete time
More informationBahadur representations for bootstrap quantiles 1
Bahadur representations for bootstrap quantiles 1 Yijun Zuo Department of Statistics and Probability, Michigan State University East Lansing, MI 48824, USA zuo@msu.edu 1 Research partially supported by
More informationConsistent Parameter Estimation for Conditional Moment Restrictions
Consistent Parameter Estimation for Conditional Moment Restrictions Shih-Hsun Hsu Department of Economics National Taiwan University Chung-Ming Kuan Institute of Economics Academia Sinica Preliminary;
More informationEXTENDED TAPERED BLOCK BOOTSTRAP
Statistica Sinica 20 (2010), 000-000 EXTENDED TAPERED BLOCK BOOTSTRAP Xiaofeng Shao University of Illinois at Urbana-Champaign Abstract: We propose a new block bootstrap procedure for time series, called
More informationSingle Equation Linear GMM with Serially Correlated Moment Conditions
Single Equation Linear GMM with Serially Correlated Moment Conditions Eric Zivot October 28, 2009 Univariate Time Series Let {y t } be an ergodic-stationary time series with E[y t ]=μ and var(y t )
More informationModel Specification Testing in Nonparametric and Semiparametric Time Series Econometrics. Jiti Gao
Model Specification Testing in Nonparametric and Semiparametric Time Series Econometrics Jiti Gao Department of Statistics School of Mathematics and Statistics The University of Western Australia Crawley
More informationOn block bootstrapping areal data Introduction
On block bootstrapping areal data Nicholas Nagle Department of Geography University of Colorado UCB 260 Boulder, CO 80309-0260 Telephone: 303-492-4794 Email: nicholas.nagle@colorado.edu Introduction Inference
More informationby Ye Cai and Mototsugu Shintani
ON THE LONG-RUN VARIANCE RATIO TEST FOR A UNIT ROOT by Ye Cai and Mototsugu Shintani Working Paper No. 05-W06 March 2005 DEPARTMENT OF ECONOMICS VANDERBILT UNIVERSITY NASHVILLE, TN 37235 www.vanderbilt.edu/econ
More informationESSAYS ON TIME SERIES ECONOMETRICS. Cheol-Keun Cho
ESSAYS ON TIME SERIES ECONOMETRICS By Cheol-Keun Cho A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Economics Doctor of Philosophy 204
More informationThe generalized method of moments
Robert M. Kunst robert.kunst@univie.ac.at University of Vienna and Institute for Advanced Studies Vienna February 2008 Based on the book Generalized Method of Moments by Alastair R. Hall (2005), Oxford
More informationaddresses: (X. Shao). Accepted author version posted online: 25 Sep 2014.
This article was downloaded by: [University of Illinois at Urbana-Champaign] On: 01 October 2014, At: 13:11 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954
More informationInference on distributions and quantiles using a finite-sample Dirichlet process
Dirichlet IDEAL Theory/methods Simulations Inference on distributions and quantiles using a finite-sample Dirichlet process David M. Kaplan University of Missouri Matt Goldman UC San Diego Midwest Econometrics
More informationTHE ERROR IN REJECTION PROBABILITY OF SIMPLE AUTOCORRELATION ROBUST TESTS
Econometrica, Vol. 72, No. 3 (May, 2004), 937 946 THE ERROR IN REJECTION PROBABILITY OF SIMPLE AUTOCORRELATION ROBUST TESTS BY MICHAEL JANSSON 1 A new class of autocorrelation robust test statistics is
More informationTESTING FOR NORMALITY IN THE LINEAR REGRESSION MODEL: AN EMPIRICAL LIKELIHOOD RATIO TEST
Econometrics Working Paper EWP0402 ISSN 1485-6441 Department of Economics TESTING FOR NORMALITY IN THE LINEAR REGRESSION MODEL: AN EMPIRICAL LIKELIHOOD RATIO TEST Lauren Bin Dong & David E. A. Giles Department
More informationThe Number of Bootstrap Replicates in Bootstrap Dickey-Fuller Unit Root Tests
Working Paper 2013:8 Department of Statistics The Number of Bootstrap Replicates in Bootstrap Dickey-Fuller Unit Root Tests Jianxin Wei Working Paper 2013:8 June 2013 Department of Statistics Uppsala
More informationRobust Backtesting Tests for Value-at-Risk Models
Robust Backtesting Tests for Value-at-Risk Models Jose Olmo City University London (joint work with Juan Carlos Escanciano, Indiana University) Far East and South Asia Meeting of the Econometric Society
More informationEconomic modelling and forecasting
Economic modelling and forecasting 2-6 February 2015 Bank of England he generalised method of moments Ole Rummel Adviser, CCBS at the Bank of England ole.rummel@bankofengland.co.uk Outline Classical estimation
More informationUniversity of Pavia. M Estimators. Eduardo Rossi
University of Pavia M Estimators Eduardo Rossi Criterion Function A basic unifying notion is that most econometric estimators are defined as the minimizers of certain functions constructed from the sample
More informationOPTIMAL BANDWIDTH CHOICE FOR INTERVAL ESTIMATION IN GMM REGRESSION. Yixiao Sun and Peter C.B. Phillips. May 2008
OPIAL BANDWIDH CHOICE FOR INERVAL ESIAION IN G REGRESSION By Yixiao Sun and Peter C.B. Phillips ay 8 COWLES FOUNDAION DISCUSSION PAPER NO. 66 COWLES FOUNDAION FOR RESEARCH IN ECONOICS YALE UNIVERSIY Box
More informationClosest Moment Estimation under General Conditions
Closest Moment Estimation under General Conditions Chirok Han Victoria University of Wellington New Zealand Robert de Jong Ohio State University U.S.A October, 2003 Abstract This paper considers Closest
More informationLocation Properties of Point Estimators in Linear Instrumental Variables and Related Models
Location Properties of Point Estimators in Linear Instrumental Variables and Related Models Keisuke Hirano Department of Economics University of Arizona hirano@u.arizona.edu Jack R. Porter Department of
More informationlarge number of i.i.d. observations from P. For concreteness, suppose
1 Subsampling Suppose X i, i = 1,..., n is an i.i.d. sequence of random variables with distribution P. Let θ(p ) be some real-valued parameter of interest, and let ˆθ n = ˆθ n (X 1,..., X n ) be some estimate
More informationCOMPARISON OF GMM WITH SECOND-ORDER LEAST SQUARES ESTIMATION IN NONLINEAR MODELS. Abstract
Far East J. Theo. Stat. 0() (006), 179-196 COMPARISON OF GMM WITH SECOND-ORDER LEAST SQUARES ESTIMATION IN NONLINEAR MODELS Department of Statistics University of Manitoba Winnipeg, Manitoba, Canada R3T
More informationQuantile regression and heteroskedasticity
Quantile regression and heteroskedasticity José A. F. Machado J.M.C. Santos Silva June 18, 2013 Abstract This note introduces a wrapper for qreg which reports standard errors and t statistics that are
More informationwhere x i and u i are iid N (0; 1) random variates and are mutually independent, ff =0; and fi =1. ff(x i )=fl0 + fl1x i with fl0 =1. We examine the e
Inference on the Quantile Regression Process Electronic Appendix Roger Koenker and Zhijie Xiao 1 Asymptotic Critical Values Like many other Kolmogorov-Smirnov type tests (see, e.g. Andrews (1993)), the
More informationImproving linear quantile regression for
Improving linear quantile regression for replicated data arxiv:1901.0369v1 [stat.ap] 16 Jan 2019 Kaushik Jana 1 and Debasis Sengupta 2 1 Imperial College London, UK 2 Indian Statistical Institute, Kolkata,
More informationEconometrica, Vol. 69, No. 6 (November, 2001), ASYMPTOTIC OPTIMALITY OF EMPIRICAL LIKELIHOOD FOR TESTING MOMENT RESTRICTIONS
Econometrica, Vol. 69, No. 6 (November, 200), 66 672 ASYMPTOTIC OPTIMALITY OF EMPIRICAL LIKELIHOOD FOR TESTING MOMENT RESTRICTIONS By Yuichi Kitamura introduction Economic theory often provides testable
More informationLecture 11 Weak IV. Econ 715
Lecture 11 Weak IV Instrument exogeneity and instrument relevance are two crucial requirements in empirical analysis using GMM. It now appears that in many applications of GMM and IV regressions, instruments
More informationChoice of Spectral Density Estimator in Ng-Perron Test: Comparative Analysis
MPRA Munich Personal RePEc Archive Choice of Spectral Density Estimator in Ng-Perron Test: Comparative Analysis Muhammad Irfan Malik and Atiq-ur- Rehman International Institute of Islamic Economics, International
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 informationAccurate Asymptotic Approximation in the Optimal GMM Frame. Stochastic Volatility Models
Accurate Asymptotic Approximation in the Optimal GMM Framework with Application to Stochastic Volatility Models Yixiao Sun UC San Diego January 21, 2014 GMM Estimation Optimal Weighting Matrix Two-step
More informationA Recursive Formula for the Kaplan-Meier Estimator with Mean Constraints
Noname manuscript No. (will be inserted by the editor) A Recursive Formula for the Kaplan-Meier Estimator with Mean Constraints Mai Zhou Yifan Yang Received: date / Accepted: date Abstract In this note
More informationDefect Detection using Nonparametric Regression
Defect Detection using Nonparametric Regression Siana Halim Industrial Engineering Department-Petra Christian University Siwalankerto 121-131 Surabaya- Indonesia halim@petra.ac.id Abstract: To compare
More informationClosest Moment Estimation under General Conditions
Closest Moment Estimation under General Conditions Chirok Han and Robert de Jong January 28, 2002 Abstract This paper considers Closest Moment (CM) estimation with a general distance function, and avoids
More informationDiagnostic Test for GARCH Models Based on Absolute Residual Autocorrelations
Diagnostic Test for GARCH Models Based on Absolute Residual Autocorrelations Farhat Iqbal Department of Statistics, University of Balochistan Quetta-Pakistan farhatiqb@gmail.com Abstract In this paper
More informationResearch Article Least Squares Estimators for Unit Root Processes with Locally Stationary Disturbance
Advances in Decision Sciences Volume, Article ID 893497, 6 pages doi:.55//893497 Research Article Least Squares Estimators for Unit Root Processes with Locally Stationary Disturbance Junichi Hirukawa and
More informationA CENTRAL LIMIT THEOREM FOR NESTED OR SLICED LATIN HYPERCUBE DESIGNS
Statistica Sinica 26 (2016), 1117-1128 doi:http://dx.doi.org/10.5705/ss.202015.0240 A CENTRAL LIMIT THEOREM FOR NESTED OR SLICED LATIN HYPERCUBE DESIGNS Xu He and Peter Z. G. Qian Chinese Academy of Sciences
More informationQuantile Regression for Residual Life and Empirical Likelihood
Quantile Regression for Residual Life and Empirical Likelihood Mai Zhou email: mai@ms.uky.edu Department of Statistics, University of Kentucky, Lexington, KY 40506-0027, USA Jong-Hyeon Jeong email: jeong@nsabp.pitt.edu
More informationHAC Corrections for Strongly Autocorrelated Time Series
HAC Corrections for Strongly Autocorrelated Time Series Ulrich K. Müller Princeton University Department of Economics Princeton, NJ, 08544 December 2012 Abstract Applied work routinely relies on heteroskedasticity
More informationFinite Sample Properties of the Dependent Bootstrap for Conditional Moment Models
1 Finite Sample Properties of the Dependent Bootstrap for Conditional Moment Models Rachida Ouysse September 5, 27 Abstract This paper assesses the finite sample refinements of the block bootstrap and
More informationAN EMPIRICAL LIKELIHOOD RATIO TEST FOR NORMALITY
Econometrics Working Paper EWP0401 ISSN 1485-6441 Department of Economics AN EMPIRICAL LIKELIHOOD RATIO TEST FOR NORMALITY Lauren Bin Dong & David E. A. Giles Department of Economics, University of Victoria
More informationGMM Estimation and Testing
GMM Estimation and Testing Whitney Newey July 2007 Idea: Estimate parameters by setting sample moments to be close to population counterpart. Definitions: β : p 1 parameter vector, with true value β 0.
More informationNon-Parametric Dependent Data Bootstrap for Conditional Moment Models
Non-Parametric Dependent Data Bootstrap for Conditional Moment Models Bruce E. Hansen University of Wisconsin September 1999 Preliminary and Incomplete Abstract A new non-parametric bootstrap is introduced
More informationNUCLEAR NORM PENALIZED ESTIMATION OF INTERACTIVE FIXED EFFECT MODELS. Incomplete and Work in Progress. 1. Introduction
NUCLEAR NORM PENALIZED ESTIMATION OF IERACTIVE FIXED EFFECT MODELS HYUNGSIK ROGER MOON AND MARTIN WEIDNER Incomplete and Work in Progress. Introduction Interactive fixed effects panel regression models
More information