Robust Backtesting Tests for Value-at-Risk Models
|
|
- Verity Webster
- 5 years ago
- Views:
Transcription
1 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
2 Outline Differences between a specification test for the quantile process and backtesting. Model risk in unconditional and independence Backtesting Exercises A Subsampling solution Monte Carlo analysis (Asymptotic theory vs Block bootstrap and Subsampling) Application to risk management in financial data Conclusions Far East and South Asia Meeting of the Econometric Society
3 Differences between Specification Tests and Backtesting Value-at-Risk: We say that m α (W t 1, θ) is a correctly specified α-th conditional VaR of Y t given W t 1 = {Y s, Z s} t 1 s=t h, h <, if and only if P (Y t m α (W t 1, θ 0 ) F t 1 ) = α, almost surely (a.s.), α (0, 1), t Z, (1) for some unknown parameter θ 0 Θ, with Θ a compact set in an Euclidean space R p and F t 1 denoting the set of available information. This condition is equivalent to the following in terms of conditional expectations: E[I t,α (θ 0 ) F t 1 ] = α, almost surely (a.s.), α (0, 1), t Z, (2) with I t,α (θ) := 1(Y t m α (W t 1, θ)) the so called hits or exceedances associated to the model m α (W t 1, θ 0 ). Far East and South Asia Meeting of the Econometric Society
4 This condition implies the so-called joint hypothesis (cf. Christoffersen (1998)), {I t,α (θ 0 )} are iid Ber(α) random variables (r.v.) for some θ 0 Θ, (3) where Ber(α) stands for a Bernoulli r.v. with parameter α. In the risk management literature Backtesting is identified with this joint hypothesis: Unconditional Backtesting test: E[I t,α (θ 0 )] = α, (4) and Independence Backtesting test: {I t,α (θ 0 )} being iid. (5) Motivation: The joint test E[I t,α (θ 0 ) F t 1 ] = α and the two tests above are not equivalent. In fact, the latter two tests do not imply necessarily the former. Far East and South Asia Meeting of the Econometric Society
5 Example 1. (Unconditional coverage test): Suppose that the true model for conditional VaR is with F Wt 1 (x) := P {Y t x I t 1 }. m α (W t 1, θ 0 ) = F 1 W t 1 (α), (6) The researcher proposes, instead, an unconditional quantile: m α (W t 1, θ 0 ) = F 1 (α), for all t, (7) with F 1 ( ) the inverse function of the unconditional distribution function F. Note from these equations that E[1(Y t m α (W t 1, θ))] = F (F 1 (α)) = α. Far East and South Asia Meeting of the Econometric Society
6 However, the unconditional quantile is misspecified to assess the conditional quantile unless F 1 W t 1 (α) = F 1 (α) a.s. Note: The popular Historical simulation (HS) VaR falls in this group of unconditional risk models. Note that in the HS case the unconditional VaR quantile is estimated nonparametrically from the empirical distribution function F n : m α (W t 1, θ 0 ) = F 1 n (α), for all t, (8) and therefore implies that this method exhibits model risk effects. Message: This method to quantify risk based on the unconditional α quantile can be successful in passing the backtesting unconditional coverage test, being however a wrong specification of the true conditional quantile process. Far East and South Asia Meeting of the Econometric Society
7 Example 2. (Independence test): Suppose that {Y t } is an autoregressive model given by Y t = ρ 1 Y t 1 + ε t, (9) with ρ 1 < 1, and ε t an error term serially independent that follows a distribution function F ε ( ). Then m α (W t 1, θ 0 ) = ρ 1 Y t 1 + Fε 1 (α), with Fε 1 (α) the α quantile of F ε. Suppose now that the researcher assumes incorrectly an alternative autoregressive process where the error term follows a Gaussian distribution function (Φ( )). The conditional (misspecified) quantile process is with Φ 1 ε (α) Fε 1 (α). m α (W t 1, θ 0 ) = ρ 1 Y t 1 + Φ 1 ε (α), Far East and South Asia Meeting of the Econometric Society
8 Note that although the risk model is wrong: m α (W t 1, θ 0 ) m α (W t 1, θ 0 ), the sequence of hits associated to m α (W t 1, θ 0 ) are iid since P {ε t Φ 1 ε (α), ε t 1 Φ 1 ε by independence of the error term. (α)} = P {ε t Φ 1 ε (α)}p {ε t 1 Φ 1 ε (α)}, Far East and South Asia Meeting of the Econometric Society
9 Implementation of Backtesting The test statistic introduced by Kupiec (1995) for the unconditional coverage hypothesis is K P K(P, R) := 1 P n t=r+1 (I t,α (θ 0 ) α), with R being the in-sample size and P the corresponding out-of-sample testing period. In practice, however, the test statistic implemented is S P S(P, R) := 1 P n t=r+1 (I t,α ( θ t 1 ) α), (10) with θ t 1 an estimator of θ 0 satisfying certain regularity conditions (cf. A4 below). Far East and South Asia Meeting of the Econometric Society
10 Estimation and Model Risk I: Assumptions Assumption A1: {Y t, Z t } t Z is strictly stationary and strong mixing process with mixing coefficients satisfying j=1 (α(j))1 2/d <, with d > 2 as in A4. Assumption A2: The family( of distributions functions {F x,) x R dw } has Lebesgue densities {f x, x R dw } that are uniformly bounded sup x R dw,y R f x(y) C and equicontinuous: for every ɛ > 0 there exists a δ > 0 such that sup x R dw, y z δ f x(y) f x (z) ɛ. Assumption A3: The model [ m α (W t 1, θ) is continuously differentiable in θ (a.s.) g α (W t 1, θ) such that E sup θ Θ0 g α (W t 1, θ) 2] < C, for a neighborhood Θ 0 of θ 0. with derivative Assumption A4: The parameter space Θ is compact in R p. The true parameter θ 0 belongs to the interior of Θ. The estimator θ t satisfies the asymptotic expansion θ t θ 0 = H(t) + o P (1), where H(t) is a p 1 vector such that H(t) = t 1 t s=1 l(y s, W s 1, θ 0 ), R 1 t s=t R+1 l(y s, W s 1, θ 0 ) and R 1 R s=1 l(y s, W s 1, θ 0 ) for the recursive, [ rolling and fixed schemes, respectively. Moreover, l(y t, W t 1, θ) is continuous (a.s.) in θ in Θ 0 and E l(y t, W t 1, θ 0 ) d] < for some d > 2. Assumption A5: R, P as n, and lim n P/R = π, 0 π <. Far East and South Asia Meeting of the Econometric Society
11 Estimation and Model Risk II: Decomposition Under the above assumptions Escanciano and Olmo (2008) showed that S P = 1 P n t=r+1 [ It,α (θ 0 ) F Wt 1 (m α (W t 1, θ 0 )) ] (11) +E [ g α(w t 1, θ 0 )f Wt 1 (m α (W t 1, θ 0 )) ] 1 H(t 1) P t=r+1 }{{} n Estimation Risk + 1 [ FWt 1 (m α (W t 1, θ 0 )) α ] P t=r+1 }{{} Model Risk n + o P (1), where g α (W t 1, θ) is the derivative of m α (W t 1, θ) with respect to θ. Far East and South Asia Meeting of the Econometric Society
12 Estimation and Model Risk III: Theorem Theorem 1: Under Assumptions A1-A5 and E[I t,α (θ 0 )] = α, S P d N(0, σ 2 a), with σa 2 = j= E[a ta t j ] + λ al (A j= E[a tl t j ] + j= E[a tl t j ] A ) + λ ll A j= E[l tl t j ]A, where Scheme λ al Recursive 1 π 1 ln(1 + π) λ ll 2 [ 1 π 1 ln(1 + π) ] Rolling, π 1 π/2 π π 2 /3 Rolling, 1 < π < 1 (2π) 1 1 (3π) 1 Fixed 0 π (12) and A = E [ g α(w t 1, θ 0 )f Wt 1 (m α (W t 1, θ 0 )) ] 1 P n t=r+1 H(t 1) and l(y s, W s 1, θ 0 ) is the score function. Far East and South Asia Meeting of the Econometric Society
13 Independence Tests I: test statistics To complement backtesting exercises one can be interested in testing the following {I t,α (θ 0 )} n t=r+1 are iid. (13) Since for Bernoulli random variables serial independence is equivalent to serial uncorrelation, it is natural to build a test for (13) based on the autocovariances γ j = Cov(I t,α (θ 0 ), I t j,α (θ 0 )) j 1. (14) They can be consistently estimated (under E[I t,α (θ 0 )] = α) by either of these tests: (Joint Test) γ P,j = 1 P j n t=r+j+1 (I t,α (θ 0 ) α)(i t j,α (θ 0 ) α) j 1. Far East and South Asia Meeting of the Econometric Society
14 (Marginal Test) ζ P,j = 1 P j n t=r+j+1 where, for θ Θ, E n [I t,α (θ)] = 1 P j (I t,α (θ 0 ) E n [I t,α (θ 0 )])(I t j,α (θ 0 ) E n [I t j,α (θ 0 )]), { n t=r+j+1 I t,α (θ) In practice, however, joint or marginal tests for (13) need to be based on estimates of the relevant parameters, such as }. or γ P,j = 1 P j n t=r+j+1 (I t,α ( θ t 1 ) α)(i t j,α ( θ t j 1 ) α) ζ P,j = 1 P j n t=r+j+1 (I t,α ( θ t 1 ) E n [I t,α ( θ t 1 )])(I t j,α ( θ t j 1 ) E n [I t j,α ( θ t j 1 )]). Far East and South Asia Meeting of the Econometric Society
15 Independence Tests II: Theorem Theorem 2: Under Assumptions A1-A5 and the joint test, for any j 1, γ P,j d N(0, σ 2 b ), where σ 2 b = E[b2 t] + λ al (B j= E[b tl t j ] + j= E[b tl t j ] B ) + λ ll B j= E[l tl t j ]B, and B = E [ g α(w t 1, θ 0 )f Wt 1 (m α (W t 1, θ 0 )){I t j,α (θ 0 ) + α} ] and b t = (I t,α (θ) α)(i t j,α (θ) α). If instead of the joint hypothesis, only (13) holds, then ζ P,j d N(0, σ 2 c), where σ 2 c = E[c 2 t] + λ al (C j= E[c tl t j ] + j= E[c tl t j ] C ) + λ ll C j= E[l tl t j ]C, with C = E [ g α(w t 1, θ 0 )f Wt 1 (m α (W t 1, θ 0 )){I t j,α (θ 0 ) + E[I t,α (θ 0 )]} ] and c t = (I t,α (θ) E[I t,α (θ)])(i t j,α (θ) E[I t j,α (θ)]). Far East and South Asia Meeting of the Econometric Society
16 Subsampling approximation: Known parameters Let T P (θ 0 ) = T P (X R+1,..., X n ; θ 0 ), be the relevant test statistic and G P (w) the corresponding cumulative distribution function: G P (w) = P (T P (θ 0 ) w). Let T b,i (θ 0 ) = T b (X i,..., X i+b 1 ; θ 0 ), i = 1,..., n b+1, be the test statistic computed with the subsample (X i,..., X i+b 1 ) of size b. Subsampling: One can approximate the sampling distribution G P (w) using the distribution of the values of T b,i (θ 0 ) computed over the n b+1 different consecutive subsamples of size b. This subsampling distribution is defined by G P,b (w) = 1 n b + 1 n b+1 i=1 1(T b,i (θ 0 ) w), w [0, ). and let c P,1 τ,b be the (1 τ)-th sample quantile of G P,b (w), i.e., c P,1 τ,b = inf{w : G P,b (w) 1 τ}. Far East and South Asia Meeting of the Econometric Society
17 Subsampling approximation: Theorem Theorem 3: Assume A1-A5 and that b/p 0 with b as P. Then, for T P (θ 0 ) any of the tests statistics K P, γ P,j or ζ P,j, we have that (i) under the joint null hypothesis, c P,1 τ,b P c1 τ and P (T P (θ 0 ) > c P,1 τ,b ) τ; (ii) under the marginal null hypothesis of unconditional coverage probability, c P,1 τ,b P c1 τ and (iii) under any fixed alternative hypothesis, P (T P (θ 0 ) > c P,1 τ,b ) τ; P (T P (θ 0 ) > c P,1 τ,b ) 1. Far East and South Asia Meeting of the Econometric Society
18 Subsampling approximation: Unknown parameters The subsampling distribution of the relevant test statistics, now S P, γ P,j or ζ P,j, is G P,b(w) = 1 n b + 1 n b+1 i=1 1( T b,i w), w [0, ), with T b,i the subsample version of the test statistic, computed with data X i,..., X i+b 1. Under the presence of estimation risk we divide each subsample X i,..., X i+b 1 in insample and out-of-sample observations according to the ratio π used for computing the original test: R b observations are used for in-sample and P b observations for out-of-sample, such that R b + P b = b and lim b P b /R b = π. The computation of T b,i is done as with the original test, say T P, but replacing the original sample by X i,..., X i+b 1 and R and P by R b and P b. Far East and South Asia Meeting of the Econometric Society
19 Theorem 4: Assume A1-A5 and that b/p 0 with b as P. Moreover, assume that lim b P b /R b = lim n P/R = π. Then, for T P any of the tests statistics S P, γ P,j or ζ P,j, and, we have that (i) under the joint null hypothesis, c P,1 τ,b P c 1 τ and P ( T P > c P,1 τ,b ) τ; (ii) under the marginal null hypothesis of unconditional coverage probability, c P,1 τ,b P c 1 τ and (iii) under any fixed alternative hypothesis, P ( T P > c P,1 τ,b ) τ; P ( T P > c P,1 τ,b ) 1. Far East and South Asia Meeting of the Econometric Society
20 Monte Carlo Experiment: Unconditional Backtesting test Data generating process: Y t = σ t ε t, with ε t N(0, 1) Φ( ), and where β 0 = 0.05, β 1 = 0.10 and β 2 = σ 2 t = β 0 + β 1 Y 2 t 1 + β 2 σ 2 t 1, (15) Model 1: True conditional VaR process (Correct specification). m α,1 (W t 1, β) = σ t Φ 1 (α), (16) with Φ 1 (α) the α quantile of the standard Gaussian distribution function. Comments: The asymptotic distribution of the unconditional coverage test is N(0, α(1 α)). Subsampling and block bootstrap methods are in this case valid alternatives to approximate the finite-sample distribution of the test statistic. Far East and South Asia Meeting of the Econometric Society
21 Model 2 : Estimated version of the true VaR process. m α,2 (W t 1, β) = σ t Φ 1 (α), (17) with β = ( β 0, β 1, β 2 ), and σ 2 t = β 0 + β 1 Y 2 t 1 + β 2 σ 2 t 1 the estimated parameters obtained from the in-sample size (R=1000 in-sample size to estimate the process, and P=500 out-of-sample evaluation period). Problem: Estimation effects. The asymptotic distribution is Gaussian but the variance is different from α(1 α). It depends on a nonlinear manner on the asymptotic ratio P/R and the values of the parameter vector β. Solutions: Subsampling and block bootstrap versions of the test account for estimation effects. Far East and South Asia Meeting of the Econometric Society
22 Model 3: Historical Simulation method. m α,3 (W t 1, β) = 1 F R (α), (18) with F R the empirical distribution function of the R = 1000 in-sample data. Problem: This method can potentially exhibit very strong estimation and model risk effects. Therefore, the asymptotic critical values of the unconditional backtesting test is not N(0, α(1 α)). The size of the test using these critical values will be very distorted under general situations. Comment: simplicity. This method is widely used by practitioners in the industry due to its Solutions: Subsampling and block bootstrap versions of the test account for estimation and model risk effects and are, in theory, able to approximate the appropriate nominal sizes. Far East and South Asia Meeting of the Econometric Society
23 Simulations of Size Φ( ) α = 0.10 α = 0.05 α = 0.01 P = 500/size K asy,1 P K bb,1 P K ss,1 P P = 500/size S asy,2 P S bb,2 P S ss,2 P P = 500/size S asy,3 P S bb,3 P S ss,3 P Table 1. Unconditional backtesting test for models (16), (17) and (18). R = 1000 and P = 500. Coverage probability α = 0.10, 0.05, b = kp 2/5 with k = Monte-Carlo replications. Far East and South Asia Meeting of the Econometric Society
24 Monte Carlo Experiment: Independence test Data generating process: Y t = σ t ε t, with ε t Φ( ), and σ 2 t = β 0 + β 1 Y 2 t 1 + β 2 σ 2 t 1, where β 0 = 0.05, β 1 = 0.10 and β 2 = Note that γ P,j is a joint test (unconditional and independence). If E[I t,α (θ 0 )] = α this test does not exhibit model neither estimation risk, its asymptotic distribution is N(0, α 2 (1 α) 2 ). ξ P,j, does not exhibit model risk, its asymptotic distribution is N(0, α 2 (1 α) 2 ). The estimated version of these test statistics, γ P,j and ξ P,j, however exhibit estimation risk. Solution: subsampling and block bootstrap methods. In the simulations below we only study γ P,1 and ξ P,1 in terms of size. Far East and South Asia Meeting of the Econometric Society
25 Simulations of Size Φ( ) α = 0.10 α = 0.05 α = 0.01 P = 500/size γ asy,1 P, γ bb,1 γ ss,1 P, P, P = 500/size ξ asy,2 P, ξ bb,2 P, ξ ss,2 P, Table 2. Empirical size for independence backtesting test for models (16) and (17). R = 1000 and P = 500. Coverage probability α = 0.10, 0.05, b = kp 2/5 with k = Monte-Carlo replications. Far East and South Asia Meeting of the Econometric Society
26 Application: Backtesting performance of VaR forecast models The models under study are Hybrid method: m α,1 (W t 1, θ t 1 ) = σ t F 1 R, ε (α), (19) with F R, ε the empirical distribution function constructed from an in-sample size of R observations, and σ t 2 = β 0 + β 1 Yt β 2 σ t 1, 2 the estimated GARCH process, where ( β 0, β 1, β 2 ) is the vector of estimates of the parameter vector (β 0, β 1, β 2 ) and { ε t } is the residual sequence. Far East and South Asia Meeting of the Econometric Society
27 Gaussian GARCH model: m α,2 (W t 1, θ t 1 ) = σ t Φ 1 (α). (20) Student-t GARCH model: m α,3 (W t 1, θ t 1 ) = σ t t 1 ν (α). (21) Riskmetrics model: m α,4 (W t 1, θ t 1 ) = σ t Rm Φ 1 (α), (22) with ( σ t Rm ) 2 = β 1 Yt (1 β 1 )( σ t 1) Rm 2. Far East and South Asia Meeting of the Econometric Society
28 Design of Backtesting Exercise A backtesting exercise for daily returns on Dow Jones Industrial Average Index over the period 02/01/1998 until 23/05/2008 containing 2500 observations. Use a fixed forecasting scheme for estimating the relevant empirical distribution functions F R and the GARCH parameters, and where the in-sample size R = 1000 is considerably greater than the out-of-sample size P = 500. The choice of this sample size is a compromise between absence of estimation risk effects (P/R < 1) and meaningful results of the subsampling and asymptotic tests (P sufficiently large). We repeat this experiment using all the information of the time series available considering rolling windows of 250 observations and giving a total of five different periods where computing backtesting tests. The rejection regions at 5% considered for the unconditional coverage and independence tests are those determined by the corresponding asymptotic normal distributions and the alternative subsampling approximations. Far East and South Asia Meeting of the Econometric Society
29 Unconditional test: Hybrid Method Backtesting test Sn at 5% for HS GARCH. k=8. Sn Subsampling Asymptotic out of sample rolling period Far East and South Asia Meeting of the Econometric Society
30 Unconditional test: Gaussian GARCH(1,1) Method Backtesting test Sn at 5% for Gaussian GARCH. k=8. Sn Subsampling Asymptotic out of sample rolling period Far East and South Asia Meeting of the Econometric Society
31 Unconditional test: Student-t GARCH(1,1) Method Backtesting test Sn at 5% for Student GARCH. k=8. Sn Subsampling Asymptotic out of sample rolling period Far East and South Asia Meeting of the Econometric Society
32 Unconditional test: Riskmetrics Method Backtesting test Sn at 5% for Rismetrics. k=8. Sn Subsampling Asymptotic out of sample rolling period Far East and South Asia Meeting of the Econometric Society
33 Comments on the plots We observe the differences between the subsampling rejection regions and the asymptotic ones for the first three methods based on GARCH estimates, and the similarities of the Riskmetrics approach with the asymptotic intervals. Therefore, whereas we reject the VaR forecasting methods obtained from using GARCH models we do not do it if Riskmetrics is employed. The choice of the asymptotic critical values is misleading most of the times since one can accept VaR models that are wrong if no attention is paid to the misspecification effects. We observe an increase in the uncertainty in all of the unconditional backtesting tests that use subsampling. Far East and South Asia Meeting of the Econometric Society
34 Independence test: Hybrid Method Dependence backtesting test \hat{\xi}_n at 5% for HS GARCH. k=8. Sn Subsampling Asymptotic out of sample rolling period Far East and South Asia Meeting of the Econometric Society
35 Independence test: Gaussian GARCH(1,1) Method Dependence backtesting test \hat{\xi}_n at 5% for Gaussian GARCH. k=8. Sn Subsampling Asymptotic out of sample rolling period Far East and South Asia Meeting of the Econometric Society
36 Independence test: Student-t GARCH(1,1) Method Dependence backtesting test \hat{\xi}_n at 5% for Student GARCH. k=8. Sn Subsampling Asymptotic out of sample rolling period Far East and South Asia Meeting of the Econometric Society
37 Independence test: Riskmetrics Method Dependence backtesting test \hat{\xi}_n at 5% for Riskmetrics. k=8. Sn Subsampling Asymptotic out of sample rolling period Far East and South Asia Meeting of the Econometric Society
38 Conclusions Backtesting techniques are statistical tests designed, specially, to uncover an excessive risk-taking from financial institutions and measured by the number of exceedances of the VaR model under scrutiny. Econometric methods that are well specified in-sample for describing the dynamics of extreme quantiles are not necessarily those that best forecast their future dynamics. Therefore, financial institutions can choose risk models for forecasting conditional VaR that although badly specified they succeed to satisfy unconditional and independence backtesting requirements. In order to implement correctly the standard backtesting procedures one needs to incorporate in the asymptotic theory certain components accounting for possible misspecifications of the risk model. A feasible alternative to the complicated asymptotic theory is resampling methods: block bootstrap and subsampling. Far East and South Asia Meeting of the Econometric Society
39 Our simulations indicate that values of b = kp 2/5 determined by k > 5 provide valid approximations of the true sampling distributions and hence of the true rejection regions for backtesting tests. We have shown that although estimation risk can be diversified by choosing a large in-sample size relative to out-of-sample, model risk cannot. Model risk is pervasive in unconditional backtests. Far East and South Asia Meeting of the Econometric Society
ROBUST BACKTESTING TESTS FOR VALUE-AT-RISK MODELS
ROBUST BACKTESTING TESTS FOR VALUE-AT-RISK MODELS J. Carlos Escanciano Indiana University, Bloomington, IN, USA Jose Olmo City University, London, UK November 2008 Abstract Backtesting methods are statistical
More informationSPECIFICATION TESTS IN PARAMETRIC VALUE-AT-RISK MODELS
SPECIFICATION TESTS IN PARAMETRIC VALUE-AT-RISK MODELS J. Carlos Escanciano Indiana University, Bloomington, IN, USA Jose Olmo City University, London, UK Abstract One of the implications of the creation
More informationPITFALLS IN BACKTESTING HISTORICAL SIMULATION MODELS
PITFALLS IN BACKTESTING HISTORICAL SIMULATION MODELS Juan Carlos Escanciano Indiana University Pei, Pei Indiana University February 14, 2011 Abstract Historical Simulation (HS) and its variant, the Filtered
More informationPitfalls in Backtesting Historical Simulation VaR Models
CAEPR Working Paper #202-003 Pitfalls in Backtesting Historical Simulation VaR Models Juan Carlos Escanciano Indiana University Pei Pei Indiana University and Chinese Academy of Finance and Development,
More informationEvaluating Value-at-Risk models via Quantile Regression
Evaluating Value-at-Risk models via Quantile Regression Luiz Renato Lima (University of Tennessee, Knoxville) Wagner Gaglianone, Oliver Linton, Daniel Smith. NASM-2009 05/31/2009 Motivation Recent nancial
More informationBacktesting Marginal Expected Shortfall and Related Systemic Risk Measures
Backtesting Marginal Expected Shortfall and Related Systemic Risk Measures Denisa Banulescu 1 Christophe Hurlin 1 Jérémy Leymarie 1 Olivier Scaillet 2 1 University of Orleans 2 University of Geneva & Swiss
More informationESTIMATION RISK EFFECTS ON BACKTESTING FOR PARAMETRIC VALUE-AT-RISK MODELS
ESTIMATION RISK EFFECTS ON BACKTESTING FOR PARAMETRIC VALUE-AT-RISK MODELS J. Carlos Escanciano Indiana University, Bloomington, IN, USA Jose Olmo City University, London, UK This draft, March 2007 Abstract
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 informationBayesian Semiparametric GARCH Models
Bayesian Semiparametric GARCH Models Xibin (Bill) Zhang and Maxwell L. King Department of Econometrics and Business Statistics Faculty of Business and Economics xibin.zhang@monash.edu Quantitative Methods
More informationBayesian Semiparametric GARCH Models
Bayesian Semiparametric GARCH Models Xibin (Bill) Zhang and Maxwell L. King Department of Econometrics and Business Statistics Faculty of Business and Economics xibin.zhang@monash.edu Quantitative Methods
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 Analysis of Power for Some Chosen VaR Backtesting Procedures - Simulation Approach
The Analysis of Power for Some Chosen VaR Backtesting Procedures - Simulation Approach Krzysztof Piontek Department of Financial Investments and Risk Management Wroclaw University of Economics ul. Komandorska
More informationBacktesting Marginal Expected Shortfall and Related Systemic Risk Measures
and Related Systemic Risk Measures Denisa Banulescu, Christophe Hurlin, Jérémy Leymarie, Olivier Scaillet, ACPR Chair "Regulation and Systemic Risk" - March 24, 2016 Systemic risk The recent nancial crisis
More informationEconometrics of financial markets, -solutions to seminar 1. Problem 1
Econometrics of financial markets, -solutions to seminar 1. Problem 1 a) Estimate with OLS. For any regression y i α + βx i + u i for OLS to be unbiased we need cov (u i,x j )0 i, j. For the autoregressive
More informationRegression Based Expected Shortfall Backtesting
Regression Based Expected Shortfall Backtesting Sebastian Bayer and Timo Dimitriadis University of Konstanz, Department of Economics, 78457 Konstanz, Germany This Version: January 15, 2018 Abstract arxiv:1804112v1
More informationAppendix of the paper: Are interest rate options important for the assessment of interest rate risk?
Appendix of the paper: Are interest rate options important for the assessment of interest rate risk? Caio Almeida,a, José Vicente b a Graduate School of Economics, Getulio Vargas Foundation b Research
More informationBacktesting Portfolio Value-at-Risk with Estimation Risk
Backtesting ortfolio Value-at-Risk with Estimation Risk ei ei Department of Economics Indiana University Abstract Nowadays the most extensively used risk measure by financial institution is known as Value-at-Risk
More informationStudies in Nonlinear Dynamics & Econometrics
Studies in Nonlinear Dynamics & Econometrics Volume 9, Issue 2 2005 Article 4 A Note on the Hiemstra-Jones Test for Granger Non-causality Cees Diks Valentyn Panchenko University of Amsterdam, C.G.H.Diks@uva.nl
More informationBacktesting Marginal Expected Shortfall and Related Systemic Risk Measures
Backtesting Marginal Expected Shortfall and Related Systemic Risk Measures Denisa Banulescu, Christophe Hurlin, Jérémy Leymarie, Olivier Scaillet February 13, 2016 Preliminary version - Please do not cite
More informationEC2001 Econometrics 1 Dr. Jose Olmo Room D309
EC2001 Econometrics 1 Dr. Jose Olmo Room D309 J.Olmo@City.ac.uk 1 Revision of Statistical Inference 1.1 Sample, observations, population A sample is a number of observations drawn from a population. Population:
More informationSFB 823. A simple and focused backtest of value at risk. Discussion Paper. Walter Krämer, Dominik Wied
SFB 823 A simple and focused backtest of value at risk Discussion Paper Walter Krämer, Dominik Wied Nr. 17/2015 A simple and focused backtest of value at risk 1 by Walter Krämer and Dominik Wied Fakultät
More informationFall 2017 STAT 532 Homework Peter Hoff. 1. Let P be a probability measure on a collection of sets A.
1. Let P be a probability measure on a collection of sets A. (a) For each n N, let H n be a set in A such that H n H n+1. Show that P (H n ) monotonically converges to P ( k=1 H k) as n. (b) For each n
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 informationNon-Stationary Time Series and Unit Root Testing
Econometrics II Non-Stationary Time Series and Unit Root Testing Morten Nyboe Tabor Course Outline: Non-Stationary Time Series and Unit Root Testing 1 Stationarity and Deviation from Stationarity Trend-Stationarity
More informationGaussian kernel GARCH models
Gaussian kernel GARCH models Xibin (Bill) Zhang and Maxwell L. King Department of Econometrics and Business Statistics Faculty of Business and Economics 7 June 2013 Motivation A regression model is often
More informationDiscussion of Bootstrap prediction intervals for linear, nonlinear, and nonparametric autoregressions, by Li Pan and Dimitris Politis
Discussion of Bootstrap prediction intervals for linear, nonlinear, and nonparametric autoregressions, by Li Pan and Dimitris Politis Sílvia Gonçalves and Benoit Perron Département de sciences économiques,
More informationOn Backtesting Risk Measurement Models
On Backtesting Risk Measurement Models Hideatsu Tsukahara Department of Economics, Seijo University e-mail address: tsukahar@seijo.ac.jp 1 Introduction In general, the purpose of backtesting is twofold:
More informationRobustní monitorování stability v modelu CAPM
Robustní monitorování stability v modelu CAPM Ondřej Chochola, Marie Hušková, Zuzana Prášková (MFF UK) Josef Steinebach (University of Cologne) ROBUST 2012, Němčičky, 10.-14.9. 2012 Contents Introduction
More informationEcon 423 Lecture Notes: Additional Topics in Time Series 1
Econ 423 Lecture Notes: Additional Topics in Time Series 1 John C. Chao April 25, 2017 1 These notes are based in large part on Chapter 16 of Stock and Watson (2011). They are for instructional purposes
More informationIndependent and conditionally independent counterfactual distributions
Independent and conditionally independent counterfactual distributions Marcin Wolski European Investment Bank M.Wolski@eib.org Society for Nonlinear Dynamics and Econometrics Tokyo March 19, 2018 Views
More informationDSGE Methods. Estimation of DSGE models: GMM and Indirect Inference. Willi Mutschler, M.Sc.
DSGE Methods Estimation of DSGE models: GMM and Indirect Inference Willi Mutschler, M.Sc. Institute of Econometrics and Economic Statistics University of Münster willi.mutschler@wiwi.uni-muenster.de Summer
More informationTesting Downside-Risk Efficiency Under Distress
Testing Downside-Risk Efficiency Under Distress Jesus Gonzalo Universidad Carlos III de Madrid Jose Olmo City University of London XXXIII Simposio Analisis Economico 1 Some key lines Risk vs Uncertainty.
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 informationMFE Financial Econometrics 2018 Final Exam Model Solutions
MFE Financial Econometrics 2018 Final Exam Model Solutions Tuesday 12 th March, 2019 1. If (X, ε) N (0, I 2 ) what is the distribution of Y = µ + β X + ε? Y N ( µ, β 2 + 1 ) 2. What is the Cramer-Rao lower
More informationCONTAGION VERSUS FLIGHT TO QUALITY IN FINANCIAL MARKETS
EVA IV, CONTAGION VERSUS FLIGHT TO QUALITY IN FINANCIAL MARKETS Jose Olmo Department of Economics City University, London (joint work with Jesús Gonzalo, Universidad Carlos III de Madrid) 4th Conference
More informationProf. Dr. Roland Füss Lecture Series in Applied Econometrics Summer Term Introduction to Time Series Analysis
Introduction to Time Series Analysis 1 Contents: I. Basics of Time Series Analysis... 4 I.1 Stationarity... 5 I.2 Autocorrelation Function... 9 I.3 Partial Autocorrelation Function (PACF)... 14 I.4 Transformation
More informationUsing all observations when forecasting under structural breaks
Using all observations when forecasting under structural breaks Stanislav Anatolyev New Economic School Victor Kitov Moscow State University December 2007 Abstract We extend the idea of the trade-off window
More informationQuantifying Stochastic Model Errors via Robust Optimization
Quantifying Stochastic Model Errors via Robust Optimization IPAM Workshop on Uncertainty Quantification for Multiscale Stochastic Systems and Applications Jan 19, 2016 Henry Lam Industrial & Operations
More informationECON 3150/4150, Spring term Lecture 6
ECON 3150/4150, Spring term 2013. Lecture 6 Review of theoretical statistics for econometric modelling (II) Ragnar Nymoen University of Oslo 31 January 2013 1 / 25 References to Lecture 3 and 6 Lecture
More informationWhen is a copula constant? A test for changing relationships
When is a copula constant? A test for changing relationships Fabio Busetti and Andrew Harvey Bank of Italy and University of Cambridge November 2007 usetti and Harvey (Bank of Italy and University of Cambridge)
More informationNon-Stationary Time Series and Unit Root Testing
Econometrics II Non-Stationary Time Series and Unit Root Testing Morten Nyboe Tabor Course Outline: Non-Stationary Time Series and Unit Root Testing 1 Stationarity and Deviation from Stationarity Trend-Stationarity
More informationAR, MA and ARMA models
AR, MA and AR by Hedibert Lopes P Based on Tsay s Analysis of Financial Time Series (3rd edition) P 1 Stationarity 2 3 4 5 6 7 P 8 9 10 11 Outline P Linear Time Series Analysis and Its Applications For
More informationWhite Noise Processes (Section 6.2)
White Noise Processes (Section 6.) Recall that covariance stationary processes are time series, y t, such. E(y t ) = µ for all t. Var(y t ) = σ for all t, σ < 3. Cov(y t,y t-τ ) = γ(τ) for all t and τ
More informationA simple nonparametric test for structural change in joint tail probabilities SFB 823. Discussion Paper. Walter Krämer, Maarten van Kampen
SFB 823 A simple nonparametric test for structural change in joint tail probabilities Discussion Paper Walter Krämer, Maarten van Kampen Nr. 4/2009 A simple nonparametric test for structural change in
More informationModified Variance Ratio Test for Autocorrelation in the Presence of Heteroskedasticity
The Lahore Journal of Economics 23 : 1 (Summer 2018): pp. 1 19 Modified Variance Ratio Test for Autocorrelation in the Presence of Heteroskedasticity Sohail Chand * and Nuzhat Aftab ** Abstract Given that
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 informationA Forecast Rationality Test that Allows for Loss Function Asymmetries
A Forecast Rationality Test that Allows for Loss Function Asymmetries Andrea A. Naghi First Draft: May, 2013 This Draft: August 21, 2014 Abstract In this paper, we propose a conditional moment type test
More informationExpecting the Unexpected: Uniform Quantile Regression Bands with an application to Investor Sentiments
Expecting the Unexpected: Uniform Bands with an application to Investor Sentiments Boston University November 16, 2016 Econometric Analysis of Heterogeneity in Financial Markets Using s Chapter 1: Expecting
More informationVolatility. Gerald P. Dwyer. February Clemson University
Volatility Gerald P. Dwyer Clemson University February 2016 Outline 1 Volatility Characteristics of Time Series Heteroskedasticity Simpler Estimation Strategies Exponentially Weighted Moving Average Use
More informationThe Size and Power of Four Tests for Detecting Autoregressive Conditional Heteroskedasticity in the Presence of Serial Correlation
The Size and Power of Four s for Detecting Conditional Heteroskedasticity in the Presence of Serial Correlation A. Stan Hurn Department of Economics Unversity of Melbourne Australia and A. David McDonald
More informationBootstrapping the Grainger Causality Test With Integrated Data
Bootstrapping the Grainger Causality Test With Integrated Data Richard Ti n University of Reading July 26, 2006 Abstract A Monte-carlo experiment is conducted to investigate the small sample performance
More informationNon-Stationary Time Series and Unit Root Testing
Econometrics II Non-Stationary Time Series and Unit Root Testing Morten Nyboe Tabor Course Outline: Non-Stationary Time Series and Unit Root Testing 1 Stationarity and Deviation from Stationarity Trend-Stationarity
More informationA comparison of four different block bootstrap methods
Croatian Operational Research Review 189 CRORR 5(014), 189 0 A comparison of four different block bootstrap methods Boris Radovanov 1, and Aleksandra Marcikić 1 1 Faculty of Economics Subotica, University
More informationMultivariate Time Series: VAR(p) Processes and Models
Multivariate Time Series: VAR(p) Processes and Models A VAR(p) model, for p > 0 is X t = φ 0 + Φ 1 X t 1 + + Φ p X t p + A t, where X t, φ 0, and X t i are k-vectors, Φ 1,..., Φ p are k k matrices, with
More informationDSGE-Models. Limited Information Estimation General Method of Moments and Indirect Inference
DSGE-Models General Method of Moments and Indirect Inference Dr. Andrea Beccarini Willi Mutschler, M.Sc. Institute of Econometrics and Economic Statistics University of Münster willi.mutschler@uni-muenster.de
More informationGARCH Models Estimation and Inference
GARCH Models Estimation and Inference Eduardo Rossi University of Pavia December 013 Rossi GARCH Financial Econometrics - 013 1 / 1 Likelihood function The procedure most often used in estimating θ 0 in
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 informationMarket Risk. MFM Practitioner Module: Quantitiative Risk Management. John Dodson. February 8, Market Risk. John Dodson.
MFM Practitioner Module: Quantitiative Risk Management February 8, 2017 This week s material ties together our discussion going back to the beginning of the fall term about risk measures based on the (single-period)
More informationThe Slow Convergence of OLS Estimators of α, β and Portfolio. β and Portfolio Weights under Long Memory Stochastic Volatility
The Slow Convergence of OLS Estimators of α, β and Portfolio Weights under Long Memory Stochastic Volatility New York University Stern School of Business June 21, 2018 Introduction Bivariate long memory
More informationEvaluating Forecast Models with an Exact Independence Test
Evaluating Forecast Models with an Exact Independence Test P. Araújo Santos M.I. Fraga Alves Instituto Politécnico de Santarém and CEAUL Universidade de Lisboa and CEAUL paulo.santos@esg.ipsantarem.pt
More informationSize and Power of the RESET Test as Applied to Systems of Equations: A Bootstrap Approach
Size and Power of the RESET Test as Applied to Systems of Equations: A Bootstrap Approach Ghazi Shukur Panagiotis Mantalos International Business School Department of Statistics Jönköping University Lund
More informationinterval forecasting
Interval Forecasting Based on Chapter 7 of the Time Series Forecasting by Chatfield Econometric Forecasting, January 2008 Outline 1 2 3 4 5 Terminology Interval Forecasts Density Forecast Fan Chart Most
More informationA New Test in Parametric Linear Models with Nonparametric Autoregressive Errors
A New Test in Parametric Linear Models with Nonparametric Autoregressive Errors By Jiti Gao 1 and Maxwell King The University of Western Australia and Monash University Abstract: This paper considers a
More informationBacktesting Value-at-Risk: From Dynamic Quantile to Dynamic Binary Tests
Backtesting Value-at-Risk: From Dynamic Quantile to Dynamic Binary Tests Elena-Ivona Dumitrescu, Christophe Hurlin, Vinson Pham To cite this version: Elena-Ivona Dumitrescu, Christophe Hurlin, Vinson Pham.
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 informationGARCH Models. Eduardo Rossi University of Pavia. December Rossi GARCH Financial Econometrics / 50
GARCH Models Eduardo Rossi University of Pavia December 013 Rossi GARCH Financial Econometrics - 013 1 / 50 Outline 1 Stylized Facts ARCH model: definition 3 GARCH model 4 EGARCH 5 Asymmetric Models 6
More informationLecture 2: Univariate Time Series
Lecture 2: Univariate Time Series Analysis: Conditional and Unconditional Densities, Stationarity, ARMA Processes Prof. Massimo Guidolin 20192 Financial Econometrics Spring/Winter 2017 Overview Motivation:
More informationAssessing financial model risk
Assessing financial model risk and an application to electricity prices Giacomo Scandolo University of Florence giacomo.scandolo@unifi.it joint works with Pauline Barrieu (LSE) and Angelica Gianfreda (LBS)
More informationFinancial Econometrics
Financial Econometrics Nonlinear time series analysis Gerald P. Dwyer Trinity College, Dublin January 2016 Outline 1 Nonlinearity Does nonlinearity matter? Nonlinear models Tests for nonlinearity Forecasting
More informationSteven Cook University of Wales Swansea. Abstract
On the finite sample power of modified Dickey Fuller tests: The role of the initial condition Steven Cook University of Wales Swansea Abstract The relationship between the initial condition of time series
More informationRobust Testing and Variable Selection for High-Dimensional Time Series
Robust Testing and Variable Selection for High-Dimensional Time Series Ruey S. Tsay Booth School of Business, University of Chicago May, 2017 Ruey S. Tsay HTS 1 / 36 Outline 1 Focus on high-dimensional
More informationPredicting bond returns using the output gap in expansions and recessions
Erasmus university Rotterdam Erasmus school of economics Bachelor Thesis Quantitative finance Predicting bond returns using the output gap in expansions and recessions Author: Martijn Eertman Studentnumber:
More informationNegative Association, Ordering and Convergence of Resampling Methods
Negative Association, Ordering and Convergence of Resampling Methods Nicolas Chopin ENSAE, Paristech (Joint work with Mathieu Gerber and Nick Whiteley, University of Bristol) Resampling schemes: Informal
More informationHeteroskedasticity in Time Series
Heteroskedasticity in Time Series Figure: Time Series of Daily NYSE Returns. 206 / 285 Key Fact 1: Stock Returns are Approximately Serially Uncorrelated Figure: Correlogram of Daily Stock Market Returns.
More informationProblem Set 2: Box-Jenkins methodology
Problem Set : Box-Jenkins methodology 1) For an AR1) process we have: γ0) = σ ε 1 φ σ ε γ0) = 1 φ Hence, For a MA1) process, p lim R = φ γ0) = 1 + θ )σ ε σ ε 1 = γ0) 1 + θ Therefore, p lim R = 1 1 1 +
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 informationBacktesting value-at-risk accuracy: a simple new test
First proof Typesetter: RH 7 December 2006 Backtesting value-at-risk accuracy: a simple new test Christophe Hurlin LEO, University of Orléans, Rue de Blois, BP 6739, 45067 Orléans Cedex 2, France Sessi
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 informationWeek 5 Quantitative Analysis of Financial Markets Characterizing Cycles
Week 5 Quantitative Analysis of Financial Markets Characterizing Cycles Christopher Ting http://www.mysmu.edu/faculty/christophert/ Christopher Ting : christopherting@smu.edu.sg : 6828 0364 : LKCSB 5036
More informationTime Series Models for Measuring Market Risk
Time Series Models for Measuring Market Risk José Miguel Hernández Lobato Universidad Autónoma de Madrid, Computer Science Department June 28, 2007 1/ 32 Outline 1 Introduction 2 Competitive and collaborative
More informationVolume 03, Issue 6. Comparison of Panel Cointegration Tests
Volume 03, Issue 6 Comparison of Panel Cointegration Tests Deniz Dilan Karaman Örsal Humboldt University Berlin Abstract The main aim of this paper is to compare the size and size-adjusted power properties
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 informationLarge sample distribution for fully functional periodicity tests
Large sample distribution for fully functional periodicity tests Siegfried Hörmann Institute for Statistics Graz University of Technology Based on joint work with Piotr Kokoszka (Colorado State) and Gilles
More informationDo Markov-Switching Models Capture Nonlinearities in the Data? Tests using Nonparametric Methods
Do Markov-Switching Models Capture Nonlinearities in the Data? Tests using Nonparametric Methods Robert V. Breunig Centre for Economic Policy Research, Research School of Social Sciences and School of
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 informationFinancial Econometrics and Quantitative Risk Managenent Return Properties
Financial Econometrics and Quantitative Risk Managenent Return Properties Eric Zivot Updated: April 1, 2013 Lecture Outline Course introduction Return definitions Empirical properties of returns Reading
More informationHypothesis Testing (May 30, 2016)
Ch. 5 Hypothesis Testing (May 30, 2016) 1 Introduction Inference, so far as we have seen, often take the form of numerical estimates, either as single points as confidence intervals. But not always. In
More informationMonitoring Forecasting Performance
Monitoring Forecasting Performance Identifying when and why return prediction models work Allan Timmermann and Yinchu Zhu University of California, San Diego June 21, 2015 Outline Testing for time-varying
More informationSequential Implementation of Monte Carlo Tests with Uniformly Bounded Resampling Risk
Sequential Implementation of Monte Carlo Tests with Uniformly Bounded Resampling Risk Axel Gandy Department of Mathematics Imperial College London a.gandy@imperial.ac.uk user! 2009, Rennes July 8-10, 2009
More informationEconometric Forecasting
Robert M. Kunst robert.kunst@univie.ac.at University of Vienna and Institute for Advanced Studies Vienna October 1, 2014 Outline Introduction Model-free extrapolation Univariate time-series models Trend
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 informationBootstrap Testing in Econometrics
Presented May 29, 1999 at the CEA Annual Meeting Bootstrap Testing in Econometrics James G MacKinnon Queen s University at Kingston Introduction: Economists routinely compute test statistics of which the
More informationDetermining and Forecasting High-Frequency Value-at-Risk by Using Lévy Processes
Determining and Forecasting High-Frequency Value-at-Risk by Using Lévy Processes W ei Sun 1, Svetlozar Rachev 1,2, F rank J. F abozzi 3 1 Institute of Statistics and Mathematical Economics, University
More informationRecitation 2: Probability
Recitation 2: Probability Colin White, Kenny Marino January 23, 2018 Outline Facts about sets Definitions and facts about probability Random Variables and Joint Distributions Characteristics of distributions
More informationAdvanced Statistics II: Non Parametric Tests
Advanced Statistics II: Non Parametric Tests Aurélien Garivier ParisTech February 27, 2011 Outline Fitting a distribution Rank Tests for the comparison of two samples Two unrelated samples: Mann-Whitney
More informationKriging models with Gaussian processes - covariance function estimation and impact of spatial sampling
Kriging models with Gaussian processes - covariance function estimation and impact of spatial sampling François Bachoc former PhD advisor: Josselin Garnier former CEA advisor: Jean-Marc Martinez Department
More informationEconometrics II - EXAM Answer each question in separate sheets in three hours
Econometrics II - EXAM Answer each question in separate sheets in three hours. Let u and u be jointly Gaussian and independent of z in all the equations. a Investigate the identification of the following
More informationEvaluating Interval Forecasts
Evaluating Interval Forecasts By Peter F. Christoffersen 1 A complete theory for evaluating interval forecasts has not been worked out to date. Most of the literature implicitly assumes homoskedastic errors
More informationInference in VARs with Conditional Heteroskedasticity of Unknown Form
Inference in VARs with Conditional Heteroskedasticity of Unknown Form Ralf Brüggemann a Carsten Jentsch b Carsten Trenkler c University of Konstanz University of Mannheim University of Mannheim IAB Nuremberg
More informationModeling conditional distributions with mixture models: Theory and Inference
Modeling conditional distributions with mixture models: Theory and Inference John Geweke University of Iowa, USA Journal of Applied Econometrics Invited Lecture Università di Venezia Italia June 2, 2005
More information