TIME SERIES DATA ANALYSIS USING EVIEWS

I Gusti Ngurah Agung

Graduate School Of Management Faculty Of Economics University Of Indonesia Ph.D. in Biostatistics and MSc. in Mathematical Statistics from University of North Carolina at Chapel Hill

John Wiley & Sons (Asia) Pte Ltd

Contents
Preface 1 EViews workfile and descriptive data analysis 1.1 What is the EViews workfile? 1.2 Basic options in EViews 1.3 Creating a workfile 1.3.1 Creating a workfile using EViews 5 or 6 1.3.2 Creating a workfile using EViews 4 1.4 Illustrative data analysis 1.4.1 Basic descriptive statistical summary 1.4.2 Box plots and outliers 1.4.3 Descriptive statistics by groups 1.4.4 Graphs over times 1.4.5 Means seasonal growth curve 1.4.6 Correlation matrix 1.4.7 Autocorrelation and partial autocorrelation 1.4.8 Bivariate graphical presentation with regression 1.5 Special notes and comments 1.6 Statistics as a sample space 2 Continuous growth models 2.1 Introduction 2.2 Classical growth models 2.3 Autoregressive growth models 2.3.1 First-order autoregressive growth models 2.3.2 AR(p) growth models 2.4. Residual tests 2.4.1 Hypothesis of no serial correlation 2.4.2 Hypothesis of the homogeneous residual term 2.4.3 Hypothesis of the normality assumption 2.4.4 Correlogram Q-statistic 2.5 Bounded autoregressive growth models 2.6 Lagged variables or autoregressive growth models xvii 1 1 1 3 3 3 7 7 11 11 12 15 15 17 18 19 22 25 25 25 29 29 30 32 33 34 34 35 38 41

Contents 2.6.1 The white estimation method 2.6.2 The Newey-West estimation method 2.6.3 The Akaike Information and Schwarz Criterions 2.6.4 Mixed lagged-variables autoregressive growth models 2.6.5 Serial correlation LM test for LV(2,1)_GM Polynomial growth model 2.7.1 Basic polynomial growth models 2.7.2 Special polynomial growth models Growth models with exogenous variables A Taylor series approximation model Alternative univariate growth models 2.10.1 A more general growth model 2.10.2 Translog additive growth models 2.10.3 Some comments 2.10.4 Growth model having interaction factors 2.10.5 Trigonometric growth models Multivariate growth models 2.11.1 The classical multivariate growth model 2.11.2 Modified multivariate growth models 2.11.3 AR(1) multivariate general growth models 2.11.4 The S-shape multivariate AR(1) general growth models Multivariate AR(p) GLM with trend 2.12.1 Kernel density and theoretical distribution Generalized multivariate models with trend 2.13.1 The simplest multivariate autoregressive model 2.13.2 Multivariate autoregressive model with two-way interactions 2.13.3 Multivariate autoregressive model with three-way interactions Special notes and comments 2.14.1 The true population model 2.14.2 Near singular matrix 2.14.3 'To Test or Not' the assumptions of the error terms Alternative multivariate models with trend 2.15.1 The lagged endogenous variables: first autoregressive model with trend 2.15.2 The lagged endogenous variables: first autoregressive model with exogenous variables and trend 2.15.3 The mixed lagged variables: first autoregressive model with trend Generalized multivariate models with time-related effects 42 43 44 44 48 49 49 55 56 59 60 60 60 63 64 69 70 70 74 78 79 79 88 95 95 100 102 104 104 105 107 113 113 114 115 118 121 121 121

2.7

2.8 2.9 2.10

2.11

2.12 2.13

2.14

2.15

2.16

Discontinuous growth models 3.1 Introduction 3.2 Piecewise growth models

Contents 3.2.1 Two-piece classical growth models Piecewise S-shape growth models 3.3.1 Two-piece linear growth models 3.4 Two-piece polynomial bounded growth models 3.4.1 Two-piece quadratic growth models 3.4.2 Two-piece third-degree bounded growth model 3.4.3 Two-piece generalized exponential growth model 3.5 Discontinuous translog linear AR(1) growth models 3.6 Alternative discontinuous growth models 3.7 Stability test 3.7.1 Chow's breakpoint test 3.7.2 Chow's forecast test 3.8 Generalized discontinuous models with trend 3.8.1 General two-piece univariate models with trend 3.8.2 Special notes and comments 3.8.3 General two-piece multivariate models with trend 3.9 General two-piece models with time-related effects 3.10 Multivariate models by states and time periods 3.10.1 Alternative models 3.10.2 Not recommended models 3.3 4 Seemingly causal models 4.1 Introduction 4.2 Statistical analysis based on a single time series 4.2.1 The means model 4.2.2 The cell-means models 4.2.3 The lagged-variable models 4.2.4 Autoregressive models 4.2.5 Lagged-variable autoregressive models 4.3 Bivariate seemingly causal models 4.3.1 The simplest seemingly causal models 4.3.2 Simplest models in three-dimensional space 4.3.3 General univariate UVAR(p,q) seemingly causal model 4.4 Trivariate seemingly causal models 4.4.1 Simple models in three-dimensional space 4.4.2 General LVAR(p,q) with exogenous variables 4.5 System equations based on trivariate time series 4.6 General system of equations 4.7 Seemingly causal models with dummy variables 4.7.1 Homogeneous time series models 4.7.2 Heterogeneous time series models 4.8 General discontinuous seemingly causal models 4.9 Additional selected seemingly causal models 4.9.1 A Third-degree polynomial function

XI

122 129 129 136 136 137 138 138 138 155 155 158 159 160 168 171 174 180 182 183 185 185 186 186 186 192 201 201 203 204 211 212 220 220 223 226 228 232 232 233 238 243 244

xii

Contents 4.9.2 A Three-dimensional bounded semilog linear model 4.9.3 Time series Cobb-Douglas models 4.9.4 Time series CES models 4.10 Final notes in developing models 4.10.1 Expert judgment 4.10.2 Other unexpected models 4.10.3 The principal component factor analysis Special cases of regression models 5.1 Introduction 5.2 Specific cases of growth curve models 5.2.1 Basic polynomial model 5.2.2 An AR(1) regression model 5.2.3 Heteroskedasticity-consistent covariance (White) 5.3 Seemingly causal models 5.3.1 Autoregressive models 5.4 Lagged variable models 5.4.1 The basic lagged-variable model 5.4.2 Some notes 5.4.3 Generalized lagged-variable autoregressive model 5.5 Cases based on the US domestic price of copper 5.5.1 Graphical representation 5.5.2 Seemingly causal model 5.5.3 Generalized translog linear model 5.5.4 Constant elasticity of substitution models 5.5.5 Models for the first difference of an endogenous variable 5.5.6 Unexpected findings 5.5.7 Multivariate linear seemingly causal models 5.6 Return rate models 5.7 Cases based on the BASICS workfile 5.7.1 Special notes VAR and system estimation methods 6.1 Introduction 6.2 The VAR models 6.2.1 The basic VAR model 6.2.2 The VAR models with exogenous variables 6.2.3 Cases based on the demo_modified workfile 6.2.4 The VAR models with dummy variables 6.2.5 Selected VAR models based on the US domestic price of copper data 6.3 The vector error correction models 6.3.1 The basic VEC model 6.3.2 General equation of the basic VEC models 244 245 249 256 256 256 257 259 259 259 260 262 262 264 265 275 275 282 282 290 291 293 296 300 304 306 310 311 314 317 319 319 320 321 323 323 341 344 354 354 360

Contents 6.3.3 The VEC models with exogenous variables 6.3.4 Some notes and comments Special notes and comments

xiii

6.4

361 366 380 381 381 383 388 389 391 392 395 400 401 405 406 406 409 417 419 419 419 420 420 424 424 424 425 429 436 436 437 438 439 440 441 441 442 442 446

Instrumental variables models 7.1 Introduction 7.2 Should we apply instrumental models? 7.3 Residual analysis in developing instrumental models 7.3.1 Testing an hypothesis corresponding to the instrumental models 7.3.2 Graphical representation of the residual series 7.4 System equation with instrumental variables 7.5 Selected cases based on the US_DPOC data 7.6 Instrumental models with time-related effects 7.7 Instrumental seemingly causal models 7.7.1 Special notes and comments 7.8 Multivariate instrumental models based on the US_DPOC 7.8.1 Simple multivariate instrumental models 7.8.2 Multivariate instrumental models 7.9 Further extension of the instrumental models ARCH models 8.1 Introduction 8.2 Options of ARCH models 8.3 Simple ARCH models 8.3.1 Simple ARCH models 8.3.2 Special notes on the ARCH models 8.4 ARCH models with exogenous variables 8.4.1 ARCH models with one exogenous variable 8.4.2 ARCH models with two exogenous variables 8.4.3 Advanced ARCH models 8.5 Alternative GARCH variance series 8.5.1 General GARCH variance series for the GARCH/TARCH model 8.5.2 General GARCH variance series for the EGARCH model 8.5.3 General GARCH variance series for the PARCH model 8.5.4 General GARCH variance series for the component ARCH(1,1) model 8.5.5 Special notes on the GARCH variance series Additional testing hypotheses 9.1 Introduction 9.2 The unit root tests 9.2.1 Simple unit root test 9.2.2 Unit root test for higher-order serial correlation

xiv 9.2.3 Comments on the unit root tests The omitted variables tests Redundant variables test (RV-test) Nonnested test (NN-test) The Ramsey RESET test Illustrative examples based on the Demo.wfl

Contents 447 448 454 456 459 461 469 469 471 473 474 477 479 491 493 494 494 497 503 503 504 504 506 508 509 509 513 514 516 516 517 518 522 524 527 527 528 528 529 530 530 531

9.3 9.4 9.5 9.6 9.7

10 Nonlinear least squares models 10.1 Introduction 10.2 Classical growth models 10.3 Generalized Cobb-Douglas models 10.3.1 Cases based on the Demo.wfl 10.3.2 Cases based on the BASIC.wfl 10.3.3 Cases based on the US_DPOC data 10.4 Generalized CES models 10.5 Special notes and comments 10.6 Other NLS models 10.6.1 Cases based on Demo.wfl 10.6.2 Cases based on the US_DPOC data 11 Nonparametric estimation methods 11.1 What is the nonparametric data analysis 11.2 Basic moving average estimates 11.2.1 Simple moving average estimates 11.2.2 The weighted moving average estimates 11.3 Measuring the best fit model 11.4 Advanced moving average models 11.4.1 The moving average models 11.4.2 The autoregressive moving average models 11.4.3 The ARMA models with covariates 11.5 Nonparametric regression based on a time series 11.5.1 The Hardle moving average models 11.5.2 The nearest neighbor fit 11.5.3 Mathematical background of the nearest neighbor 11.6 The local polynomial Kernel fit regression 11.7 Nonparametric growth models Appendix A: Models for a single time series A.l The simplest model A. 1.1 OLS estimates A. 1.2 Properties of the error terms A. 1.3 Maximum likelihood estimates A.2 First-order autoregressive models A.2.1 Properties of the parameters A.2.2 Autocorrelation function of an AR(1) model

fit

Contents A.2.3 Estimates of the parameters A.3 Second-order autoregressive model A.3.1 Properties of the parameters A.3.2 Autocorrelation function of an AR(2) model A.3.3 Estimates of the parameters A.4 First-order moving average model A.5 Second-order moving average model A.6 The simplest ARMA model A.7 General ARMA model A.7.1 Derivation of the ACF A.7.2 Estimation method Appendix B: Simple linear models B.l The simplest linear model B.l.l Least squares estimators B.2 Linear model with basic assumptions B.2.1 Sampling distributions of the model parameters B.2.2 Student's f-statistic B.2.3 Analysis of variance table B.2.4 Coefficient of determination B.3 Maximum likelihood estimation method B.4 First-order autoregressive linear model B.4.1 Two-stage estimation method B.4.2 Durbin-Watson statistic B.4.3 Properties of the error term /j,t B.4.4 Maximum likelihood estimation method B.5 AR(p) linear model B.5.1 Estimation method B.5.2 Properties of \ix B.6 Alternative models B.6.1 Alternative 1: The simplest model with trend B.6.2 Alternative 2: The classical growth model B.6.3 Alternative 3: The AR(p) polynomial model B.6.4 Alternative 4: The AR(p) return rate model B.6.5 Alternative 5: The bounded translog linear (Cobb-Douglas) AR(p) model B.7 Lagged-variable model B.8 Lagged-variable autoregressive models B.8.1 The simplest lagged-variable autoregressive model B.8.2 General lagged-variable autoregressive model B.9 Special notes and comments Appendix C: General linear models C.l General linear model with i.i.d. Gaussian disturbances 1.1 The OLS estimates

XV

532 533 533 533 534 535 536 537 538 538 541 543 543 543 544 545 546 546 547 548 550 550 551 551 552 553 554 554 555 555 555 556 556 556 556 557 557 559 560 561 561 562

xvi

Contents . 1.2 Maximum likelihood estimates C.1.3 Student's f-statistic 1.4 The Wald form of the OLS F-test AR(1) general linear model C.2.1 Properties of /j,t C.2.2 Estimation method AR(p) general linear model General lagged-variable autoregressive model General models with Gaussian errors C.5.1 Gaussian errors with a known variance covariance matrix C.5.2 Generalized least squares with a known covariance matrix C.5.3 GLS and ML estimations C.5.4 The variance of the error is proportional to the square of one of the explanatory variables C.5.5 Generalized least squares with an unknown covariance matrix 563 564 564 565 566 566 567 567 568 568 569 570 570 571 573 573 574 575 576 576 577 577 578 579 579 580 581 581 581 582 583 584 584 585 585 585 586 587 589 593

C.2

C.3 C.4 C.5

Appendix D: Multivariate general linear models D.l Multivariate general linear models D.2 Moments of an endogenous multivariate D.3 Vector autoregressive model D.4 Vector moving average model D.5 Vector autoregressive moving average model D.6 Simple multivariate models with exogenous variables D.6.1 The simplest multivariate model D.6.2 Simple model with a multidimensional exogenous variable D.6.3 A more general model D.6.4 Selected bivariate time series models D.6.5 Bivariate Granger causality tests D.6.6 Simultaneous causal model D.6.7 Additional bivariate models D.7 General estimation methods D.7.1 The OLS estimates D.8 Maximum likelihood estimation for an MGLM D.8.1 Student's f-test D.8.2 The Wald form of the OLS F-test D.8.3 Residual analysis D.9 MGLM with autoregressive errors D.9.1 AR(p) MGLM with equal sets of exogenous variables D.9.2 AR(p) MGLM with unequal sets of exogenous variables D.9.3 Special notes and comments References Index