1 A NEW SPECIFICATION TOOL FOR STOCHASTIC VOLATILITY MODELS BASED ON THE TAYLOR EFFECT A. Pérez 1 and E. Ruiz 2 1 Dpto. Economía Aplicada Universidad de.

Slides:



Advertisements
Similar presentations
FINANCIAL TIME-SERIES ECONOMETRICS SUN LIJIAN Feb 23,2001.
Advertisements

Dynamic panels and unit roots
Copyright(© MTS-2002GG): You are free to use and modify these slides for educational purposes, but please if you improve this material send us your new.
Generalized Method of Moments: Introduction
Chapter 2. Unobserved Component models Esther Ruiz PhD Program in Business Administration and Quantitative Analysis Financial Econometrics.
A.M. Alonso, C. García-Martos, J. Rodríguez, M. J. Sánchez Seasonal dynamic factor model and bootstrap inference: Application to electricity market forecasting.
Model Building For ARIMA time series
Unit Roots & Forecasting
Time Series Building 1. Model Identification
STAT 497 APPLIED TIME SERIES ANALYSIS
IAOS 2014 Conference – Meeting the Demands of a Changing World Da Nang, Vietnam, 8-10 October 2014 ROBUST REGRESSION IMPUTATION: CONSIDERATION ON THE INFLUENCE.
BABS 502 Lecture 9 ARIMA Forecasting II March 23, 2009.
Simulation Where real stuff starts. ToC 1.What, transience, stationarity 2.How, discrete event, recurrence 3.Accuracy of output 4.Monte Carlo 5.Random.
On Threshold ARCH Models with Gram- Charlier Density Xuan Zhou and W. K. Li Department of Statistics and Actuarial Science,HKU.
Descriptive statistics Experiment  Data  Sample Statistics Experiment  Data  Sample Statistics Sample mean Sample mean Sample variance Sample variance.
Bootstrap in Finance Esther Ruiz and Maria Rosa Nieto (A. Rodríguez, J. Romo and L. Pascual) Department of Statistics UNIVERSIDAD CARLOS III DE MADRID.
Applied Geostatistics
2. Point and interval estimation Introduction Properties of estimators Finite sample size Asymptotic properties Construction methods Method of moments.
Volatility Models Fin250f: Lecture 5.2 Fall 2005 Reading: Taylor, chapter 9.
Comparative survey on non linear filtering methods : the quantization and the particle filtering approaches Afef SELLAMI Chang Young Kim.
ARIMA Forecasting Lecture 7 and 8 - March 14-16, 2011
Financial Econometrics
Deterministic Solutions Geostatistical Solutions
Economics 20 - Prof. Anderson
A First Peek at the Extremogram: a Correlogram of Extremes 1. Introduction The Autocorrelation function (ACF) is widely used as a tool for measuring Serial.
Modern methods The classical approach: MethodProsCons Time series regression Easy to implement Fairly easy to interpret Covariates may be added (normalization)
K. Ensor, STAT Spring 2005 The Basics: Outline What is a time series? What is a financial time series? What is the purpose of our analysis? Classification.
Modern methods The classical approach: MethodProsCons Time series regression Easy to implement Fairly easy to interpret Covariates may be added (normalization)
ARMA models Gloria González-Rivera University of California, Riverside
Simulation Output Analysis
STAT 497 LECTURE NOTES 2.
The Examination of Residuals. The residuals are defined as the n differences : where is an observation and is the corresponding fitted value obtained.
1 Institute of Engineering Mechanics Leopold-Franzens University Innsbruck, Austria, EU H.J. Pradlwarter and G.I. Schuëller Confidence.
Monte Carlo Simulation CWR 6536 Stochastic Subsurface Hydrology.
Detection and Estimation of long memory in the Exchange rate volatility of the peso-dolar Alejandro Fonseca EGADE Business School, Campus Monterrey
Random Regressors and Moment Based Estimation Prepared by Vera Tabakova, East Carolina University.
The Examination of Residuals. Examination of Residuals The fitting of models to data is done using an iterative approach. The first step is to fit a simple.
Explaining the statistical features of the Spanish Stock Market from the bottom-up. José A. Pascual, Javier Pajares, and Adolfo López. InSiSoc Group. Valladolid.
Various topics Petter Mostad Overview Epidemiology Study types / data types Econometrics Time series data More about sampling –Estimation.
Academy of Economic Studies DOCTORAL SCHOOL OF FINANCE AND BANKING Bucharest 2003 Long Memory in Volatility on the Romanian Stock Market Msc Student: Gabriel.
Maximum Likelihood Estimation Methods of Economic Investigation Lecture 17.
K. Ensor, STAT Spring 2005 Model selection/diagnostics Akaike’s Information Criterion (AIC) –A measure of fit plus a penalty term for the number.
Time Series Basics (2) Fin250f: Lecture 3.2 Fall 2005 Reading: Taylor, chapter , 3.9(skip 3.6.1)
Lecture 4: Statistics Review II Date: 9/5/02  Hypothesis tests: power  Estimation: likelihood, moment estimation, least square  Statistical properties.
Week 21 Stochastic Process - Introduction Stochastic processes are processes that proceed randomly in time. Rather than consider fixed random variables.
Dissertation paper Modelling and Forecasting Volatility Index based on the stochastic volatility models MSc Student: LAVINIA-ROXANA DAVID Supervisor: Professor.
Time Series Basics Fin250f: Lecture 8.1 Spring 2010 Reading: Brooks, chapter
Estimation Method of Moments (MM) Methods of Moment estimation is a general method where equations for estimating parameters are found by equating population.
Dynamic Models, Autocorrelation and Forecasting ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes.
EC208 – Introductory Econometrics. Topic: Spurious/Nonsense Regressions (as part of chapter on Dynamic Models)
Previously Definition of a stationary process (A) Constant mean (B) Constant variance (C) Constant covariance White Noise Process:Example of Stationary.
The Unscented Particle Filter 2000/09/29 이 시은. Introduction Filtering –estimate the states(parameters or hidden variable) as a set of observations becomes.
Univariate Time series - 2 Methods of Economic Investigation Lecture 19.
Correlogram - ACF. Modeling for Forecast Forecast Data The Base Model Linear Trend Logistic Growth Others Models Look for a best approximation of the.
Ka-fu Wong © 2007 ECON1003: Analysis of Economic Data Lesson0-1 Supplement 2: Comparing the two estimators of population variance by simulations.
Multiple Random Variables and Joint Distributions
Large Sample Theory EC 532 Burak Saltoğlu.
Signal processing.
Lecture 8 ARIMA Forecasting II
Econometric methods of analysis and forecasting of financial markets
Model Building For ARIMA time series
Applied Econometric Time-Series Data Analysis
Alternative Investments and Risk Measurement
Large Sample Theory EC 532 Burak Saltoğlu.
Serial Correlation and Heteroskedasticity in Time Series Regressions
Serial Correlation and Heteroscedasticity in
Testing for near integration with stationary covariates
The Examination of Residuals
Lecturer Dr. Veronika Alhanaqtah
Serial Correlation and Heteroscedasticity in
Presentation transcript:

1 A NEW SPECIFICATION TOOL FOR STOCHASTIC VOLATILITY MODELS BASED ON THE TAYLOR EFFECT A. Pérez 1 and E. Ruiz 2 1 Dpto. Economía Aplicada Universidad de Valladolid (Spain) 2 Dpto. Estadística Universidad Carlos III de Madrid (Spain) XXXIII SIMPOSIO DE ANÁLISIS ECONÓMICO XXXIII SIMPOSIO DE ANÁLISIS ECONÓMICO Zaragoza, December 12, 2008

2 Outline  Motivation and background  The Taylor property in ARSV and LMSV models  A new tool for model adequacy based on Taylor effect Asymptotic and finite sample properties  Empirical application  Conclusions

3 Motivation and background Sample autocorrelations of absolute returns are larger than those of squares; Taylor (1986) Autocorrelations of powered-absolute returns, |y t | , are highest for  1, i.e. absolute returns ; Ding et al (1993)  Taylor effect, Granger and Ding (1995) Such autocorrelations tend to persist for long lags ==> possible long-memory in volatility; Ding et al (1993)

4

5

6 Reasons for interest in autocorrelations of |y t |  Model adequacy: any theoretical model should be able to replicate these sample correlation patterns; Ding et al (1993), Baillie and Chung (2001), Karanasos et al (2004) Model selection: correlations of squares are not enough to discriminate models; Franq & Zakoian (2008) The power transformation that maximizes correlations related to predictability; Higgins & Bera (1992) Improved estimators of conditional heteroskedastic models; Deo et al (2006)

7 Taylor effect in LMSV models LMSV(1, d,0) model: y t : series of returns  t : the volatility  t ~ IID(0,1) symmetric  t ~ NID(0, ) independent of  t d <0.5, |  |<1 for stationarity Note: d =0 => ~ AR(1) => ARSV(1) model Note:  =Corr(  t,  t+1 ) + Gaussian => A-LMSV model

8 Kurtosis of y t => k y = k  exp( ) Moments and dynamic structure: ACF of | y t |  in LMSV model; Harvey (1998)  k (  )=  (| y t | ,| y t-k |  ) = where   ={E( |  t | 2  )}/{E( |  t |  )} 2   k (  ) depends on:, , d, , distribution of  t is the acf of

9

10 Let  We focus on  1 (  ) and  max (1)  1 (  )= d =0: ARSV(1) =>  h (1)=   =0: LMSV(0, d,0) =>  h (1)= d /(1- d ) Persistence

11

12 A new tool for model adequacy Asymptotic properties: Gourieroux & Jasiak (2002) Consistent estimator of  max ( k ) Asymptotically Normal:

13 Finite samples: Monte Carlo experiment 1000 series of sizes T={500, 1000, 5000}  ={0.8,0.98}, ={0.01,0.05,0.1}, d ={0.3,0.45}  t ~N(0,1) and Student t-7 A grid of values of  (0,3)   j  {0.01,0.02,…,3} For each replicate i  compute r 1i (  j ), j=1,…,300 For each  j  sample mean of r 1i (  j ) and 90%,95% For replicate i  Pick up, i =1,..,1000

14

15 Kernel densities of

16 New tool proposed Given a data set and its fitted SV model with its estimated parameters: (  =   N(0,1)) Define: = Test for model adequacy  H 0 :  max (k)=  Reject, at  %, when outside the 100(1-  )% confidence region of the asymptotic distribution

17 Summary descriptive statistics of returns* SERIE EuroBPCANYenSP500Nikkei FTSE100 IBEX35 Size Kurtosis r 1 (1) Q | Y | (50) r 1 (2) Q Y 2 (50) * F iltered by fitting MA(1) and/or correcting possible outliers >5  t/T Empirical application

18 Estimation results of ARSV(1) models SERIESEuroBPCANYenSP500Nikkei225FTSE100IBEX35     (0.005) (0.002) (0.003) (0.003) (0.002) (0.004) (0.005) (0.005) (0.001) (0.002) (0.003) (0.004) (0.004) (0.004) ∞ (0.003) (0.004) ∞   Q |  | (10) Q |  | (50) * 81.5** 22.2* 103** ** * 75.2* Q  2 (10) Q  2 (50) ** 117** ** 123** 23.9** 70.7* 43.3** 91.1** 38.9** 81.3**

19 Estimation results of LMSV models SERIESEuroBPCANYenSP500Nikkei225FTSE100IBEX35  d  ∞ ∞   Q |  | (10) Q |  | (50) ** 86.3** 28.8* 89.5** 22.4* 70.3* 45.3** 115** * 68.3* Q  2 (10) Q  2 (50) ** 71.8* * 18.9* ** 80.9** 21.1*

20

21 Conclusions 1.ARSV, LMSV and A-LMSV models are able to generate Taylor effect for the most realistic parameter sets 2.Sample and theoretical autocorrelations of | y t |  peak at similar values of  3.Use as an additional tool for model adequacy of a fitted SV model 4. consistent and asymptotically Normal