Experience Seamless Full Ebook Downloads for Every Genre at textbookfull.

com

MATLAB Econometrics Toolbox User s Guide The

Mathworks

https://textbookfull.com/product/matlab-econometrics-
toolbox-user-s-guide-the-mathworks/

OR CLICK BUTTON

DOWNLOAD NOW

Explore and download more ebook at https://textbookfull.com

 Recommended digital products (PDF, EPUB, MOBI) that
you can download immediately if you are interested.

MATLAB Bioinformatics Toolbox User s Guide The Mathworks

https://textbookfull.com/product/matlab-bioinformatics-toolbox-user-s-
guide-the-mathworks/

textboxfull.com

MATLAB Mapping Toolbox User s Guide The Mathworks

https://textbookfull.com/product/matlab-mapping-toolbox-user-s-guide-
the-mathworks/

textboxfull.com

MATLAB Optimization Toolbox User s Guide The Mathworks

https://textbookfull.com/product/matlab-optimization-toolbox-user-s-
guide-the-mathworks/

textboxfull.com

MATLAB Trading Toolbox User s Guide The Mathworks

https://textbookfull.com/product/matlab-trading-toolbox-user-s-guide-
the-mathworks/

textboxfull.com
MATLAB Computer Vision Toolbox User s Guide The Mathworks

https://textbookfull.com/product/matlab-computer-vision-toolbox-user-
s-guide-the-mathworks/

textboxfull.com

MATLAB Curve Fitting Toolbox User s Guide The Mathworks

https://textbookfull.com/product/matlab-curve-fitting-toolbox-user-s-
guide-the-mathworks/

textboxfull.com

MATLAB Fuzzy Logic Toolbox User s Guide The Mathworks

https://textbookfull.com/product/matlab-fuzzy-logic-toolbox-user-s-
guide-the-mathworks/

textboxfull.com

MATLAB Global Optimization Toolbox User s Guide The

Mathworks

https://textbookfull.com/product/matlab-global-optimization-toolbox-
user-s-guide-the-mathworks/

textboxfull.com

MATLAB Image Processing Toolbox User s Guide The Mathworks

https://textbookfull.com/product/matlab-image-processing-toolbox-user-
s-guide-the-mathworks/

textboxfull.com
Econometrics Toolbox™
User's Guide

R2020a
How to Contact MathWorks

Latest news: www.mathworks.com

Sales and services: www.mathworks.com/sales_and_services

User community: www.mathworks.com/matlabcentral

Technical support: www.mathworks.com/support/contact_us

Phone: 508-647-7000

The MathWorks, Inc.

1 Apple Hill Drive
Natick, MA 01760-2098
Econometrics Toolbox™ User's Guide
© COPYRIGHT 1999–2020 by The MathWorks, Inc.
The software described in this document is furnished under a license agreement. The software may be used or copied
only under the terms of the license agreement. No part of this manual may be photocopied or reproduced in any form
without prior written consent from The MathWorks, Inc.
FEDERAL ACQUISITION: This provision applies to all acquisitions of the Program and Documentation by, for, or through
the federal government of the United States. By accepting delivery of the Program or Documentation, the government
hereby agrees that this software or documentation qualifies as commercial computer software or commercial computer
software documentation as such terms are used or defined in FAR 12.212, DFARS Part 227.72, and DFARS 252.227-7014.
Accordingly, the terms and conditions of this Agreement and only those rights specified in this Agreement, shall pertain
to and govern the use, modification, reproduction, release, performance, display, and disclosure of the Program and
Documentation by the federal government (or other entity acquiring for or through the federal government) and shall
supersede any conflicting contractual terms or conditions. If this License fails to meet the government's needs or is
inconsistent in any respect with federal procurement law, the government agrees to return the Program and
Documentation, unused, to The MathWorks, Inc.
Trademarks
MATLAB and Simulink are registered trademarks of The MathWorks, Inc. See
www.mathworks.com/trademarks for a list of additional trademarks. Other product or brand names may be
trademarks or registered trademarks of their respective holders.
Patents
MathWorks products are protected by one or more U.S. patents. Please see www.mathworks.com/patents for
more information.
Revision History
October 2008 Online only Version 1.0 (Release 2008b)
March 2009 Online only Revised for Version 1.1 (Release 2009a)
September 2009 Online only Revised for Version 1.2 (Release 2009b)
March 2010 Online only Revised for Version 1.3 (Release 2010a)
September 2010 Online only Revised for Version 1.4 (Release 2010b)
April 2011 Online only Revised for Version 2.0 (Release 2011a)
September 2011 Online only Revised for Version 2.0.1 (Release 2011b)
March 2012 Online only Revised for Version 2.1 (Release 2012a)
September 2012 Online only Revised for Version 2.2 (Release 2012b)
March 2013 Online only Revised for Version 2.3 (Release 2013a)
September 2013 Online only Revised for Version 2.4 (Release 2013b)
March 2014 Online Only Revised for Version 3.0 (Release 2014a)
October 2014 Online Only Revised for Version 3.1 (Release 2014b)
March 2015 Online Only Revised for Version 3.2 (Release 2015a)
September 2015 Online Only Revised for Version 3.3 (Release 2015b)
March 2016 Online Only Revised for Version 3.4 (Release 2016a)
September 2016 Online Only Revised for Version 3.5 (Release 2016b)
March 2017 Online Only Revised for Version 4.0 (Release 2017a)
September 2017 Online Only Revised for Version 4.1 (Release 2017b)
March 2018 Online Only Revised for Version 5.0 (Release 2018a)
September 2018 Online Only Revised for Version 5.1 (Release 2018b)
March 2019 Online Only Revised for Version 5.2 (Release 2019a)
September 2019 Online Only Revised for Version 5.3 (Release 2019b)
March 2020 Online Only Revised for Version 5.4 (Release 2020a)
 Contents

Getting Started
1
Econometrics Toolbox Product Description . . . . . . . . . . . . . . . . . . . . . . . . 1-2

Econometric Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-3

Model Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-3
Econometrics Toolbox Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-3

Econometrics Toolbox Model Objects, Properties, and Object Functions

.......................................................... 1-7
Model Objects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-7
Model Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-8
Specify Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-10
Retrieve Model Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-13
Modify Model Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-14
Object Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-15

Stochastic Process Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-16

What Is a Stochastic Process? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-16
Stationary Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-17
Linear Time Series Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-17
Unit Root Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-18
Lag Operator Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-19
Characteristic Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-20

Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-22

Data Preprocessing
2
Data Transformations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-2
Why Transform? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-2
Common Data Transformations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-2

Trend-Stationary vs. Difference-Stationary Processes . . . . . . . . . . . . . . . . 2-6

Nonstationary Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-6
Trend Stationary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-7
Difference Stationary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-7

Specify Lag Operator Polynomials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-9

Lag Operator Polynomial of Coefficients . . . . . . . . . . . . . . . . . . . . . . . . . . 2-9
Difference Lag Operator Polynomials . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-11

iii
 Nonseasonal Differencing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-13

Nonseasonal and Seasonal Differencing . . . . . . . . . . . . . . . . . . . . . . . . . . 2-16

Time Series Decomposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-19

Moving Average Filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-21

Moving Average Trend Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-22

Parametric Trend Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-25

Hodrick-Prescott Filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-30

Using the Hodrick-Prescott Filter to Reproduce Their Original Result

......................................................... 2-31

Seasonal Filters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-35

What Is a Seasonal Filter? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-35
Stable Seasonal Filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-35
Sn × m seasonal filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-36

Seasonal Adjustment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-38

What Is Seasonal Adjustment? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-38
Deseasonalized Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-38
Seasonal Adjustment Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-38

Seasonal Adjustment Using a Stable Seasonal Filter . . . . . . . . . . . . . . . . 2-40

Seasonal Adjustment Using S(n,m) Seasonal Filters . . . . . . . . . . . . . . . . 2-46

Model Selection
3
Box-Jenkins Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-2

Box-Jenkins Model Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-4

Autocorrelation and Partial Autocorrelation . . . . . . . . . . . . . . . . . . . . . . 3-11

What Are Autocorrelation and Partial Autocorrelation? . . . . . . . . . . . . . . 3-11
Theoretical ACF and PACF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-11
Sample ACF and PACF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-11

Ljung-Box Q-Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-13

Detect Autocorrelation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-15

Compute Sample ACF and PACF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-15
Conduct the Ljung-Box Q-Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-17

Engle’s ARCH Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-20

iv Contents
Detect ARCH Effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-22
Test Autocorrelation of Squared Residuals . . . . . . . . . . . . . . . . . . . . . . . 3-22
Conduct Engle's ARCH Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-24

Unit Root Nonstationarity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-27

What Is a Unit Root Test? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-27
Modeling Unit Root Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-27
Available Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-28
Testing for Unit Roots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-30

Unit Root Tests . . . . . . . . . . . . . . . . . . ............................ 3-32

Test Simulated Data for a Unit Root ............................ 3-32
Test Time Series Data for Unit Root ............................ 3-37
Test Stock Data for a Random Walk ............................ 3-39

Assess Stationarity of a Time Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-42

Information Criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-45

Model Comparison Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-46

Available Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-46
Likelihood Ratio Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-48
Lagrange Multiplier Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-48
Wald Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-48
Covariance Matrix Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-49

Conduct Lagrange Multiplier Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-50

Conduct Wald Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-53

Compare GARCH Models Using Likelihood Ratio Test . . . . . . . . . . . . . . . 3-55

Check Fit of Multiplicative ARIMA Model . . . . . . . . . . . . . . . . . . . . . . . . . 3-58

Goodness of Fit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-63

Residual Diagnostics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-64

Check Residuals for Normality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-64
Check Residuals for Autocorrelation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-64
Check Residuals for Conditional Heteroscedasticity . . . . . . . . . . . . . . . . 3-64

Assess Predictive Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-66

Nonspherical Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-67

What Are Nonspherical Models? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-67

Plot a Confidence Band Using HAC Estimates . . . . . . . . . . . . . . . . . . . . . 3-68

Change the Bandwidth of a HAC Estimator . . . . . . . . . . . . . . . . . . . . . . . 3-75

Check Model Assumptions for Chow Test . . . . . . . . . . . . . . . . . . . . . . . . . 3-80

Power of the Chow Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-87

v
 Econometric Modeler
4
Econometric Modeler App Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-2
Prepare Data for Econometric Modeler App . . . . . . . . . . . . . . . . . . . . . . . 4-3
Import Time Series Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-3
Perform Exploratory Data Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-5
Fitting Models to Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-15
Conducting Goodness-of-Fit Checks . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-23
Finding Model with Best In-Sample Fit . . . . . . . . . . . . . . . . . . . . . . . . . . 4-29
Export Session Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-30

Specifying Lag Operator Polynomials Interactively . . . . . . . . . . . . . . . . . 4-37

Specify Lag Structure Using Lag Order Tab . . . . . . . . . . . . . . . . . . . . . . 4-38
Specify Lag Structure Using Lag Vector Tab . . . . . . . . . . . . . . . . . . . . . . 4-40

Prepare Time Series Data for Econometric Modeler App . . . . . . . . . . . . 4-43

Prepare Table of Multivariate Data for Import . . . . . . . . . . . . . . . . . . . . . 4-43
Prepare Numeric Vector for Import . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-44

Import Time Series Data into Econometric Modeler App . . . . . . . . . . . . 4-46

Import Data from MATLAB Workspace . . . . . . . . . . . . . . . . . . . . . . . . . . 4-46
Import Data from MAT-File . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-47

Plot Time Series Data Using Econometric Modeler App . . . . . . . . . . . . . 4-50

Plot Univariate Time Series Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-50
Plot Multivariate Time Series and Correlations . . . . . . . . . . . . . . . . . . . . 4-51

Detect Serial Correlation Using Econometric Modeler App . . . . . . . . . . 4-56

Plot ACF and PACF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-56
Conduct Ljung-Box Q-Test for Significant Autocorrelation . . . . . . . . . . . . 4-58

Detect ARCH Effects Using Econometric Modeler App . . . . . . . . . . . . . . 4-62

Inspect Correlograms of Squared Residuals for ARCH Effects . . . . . . . . . 4-62
Conduct Ljung-Box Q-Test on Squared Residuals . . . . . . . . . . . . . . . . . . 4-65
Conduct Engle's ARCH Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-67

Assess Stationarity of Time Series Using Econometric Modeler . . . . . . . 4-70

Test Assuming Unit Root Null Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-70
Test Assuming Stationary Null Model . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-73
Test Assuming Random Walk Null Model . . . . . . . . . . . . . . . . . . . . . . . . 4-77

Assess Collinearity Among Multiple Series Using Econometric Modeler

App . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-81

Transform Time Series Using Econometric Modeler App . . . . . . . . . . . . 4-84

Apply Log Transformation to Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-84
Stabilize Time Series Using Nonseasonal Differencing . . . . . . . . . . . . . . 4-88
Convert Prices to Returns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-91
Remove Seasonal Trend from Time Series Using Seasonal Difference . . . 4-94
Remove Deterministic Trend from Time Series . . . . . . . . . . . . . . . . . . . . 4-97

Implement Box-Jenkins Model Selection and Estimation Using

Econometric Modeler App . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-99

vi Contents
 Select ARCH Lags for GARCH Model Using Econometric Modeler App
........................................................ 4-109

Estimate Multiplicative ARIMA Model Using Econometric Modeler App

........................................................ 4-118

Perform ARIMA Model Residual Diagnostics Using Econometric Modeler

App . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-128

Specify t Innovation Distribution Using Econometric Modeler App . . . 4-137

Compare Predictive Performance After Creating Models Using

Econometric Modeler App . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-141

Estimate ARIMAX Model Using Econometric Modeler App . . . . . . . . . . 4-148

Estimate Regression Model with ARMA Errors Using Econometric

Modeler App . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-156

Compare Conditional Variance Model Fit Statistics Using Econometric

Modeler App . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-170

Perform GARCH Model Residual Diagnostics Using Econometric Modeler

App . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-179

Share Results of Econometric Modeler App Session . . . . . . . . . . . . . . . 4-186

Time Series Regression Models

5
Time Series Regression Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-3

Regression Models with Time Series Errors . . . . . . . . . . . . . . . . . . . . . . . . 5-5

What Are Regression Models with Time Series Errors? . . . . . . . . . . . . . . . 5-5
Conventions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-5

Create Regression Models with ARIMA Errors . . . . . . . . . . . . . . . . . . . . . . 5-8

Default Regression Model with ARIMA Errors Specifications . . . . . . . . . . 5-8
Specify regARIMA Models Using Name-Value Pair Arguments . . . . . . . . . . 5-9
Specify Linear Regression Models Using Econometric Modeler App . . . . 5-15

Specify the Default Regression Model with ARIMA Errors . . . . . . . . . . . 5-19

Modify regARIMA Model Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-21

Modify Properties Using Dot Notation . . . . . . . . . . . . . . . . . . . . . . . . . . 5-21
Nonmodifiable Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-23

Create Regression Models with AR Errors . . . . . . . . . . . . . . . . . . . . . . . . 5-26

Default Regression Model with AR Errors . . . . . . . . . . . . . . . . . . . . . . . . 5-26
AR Error Model Without an Intercept . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-27
AR Error Model with Nonconsecutive Lags . . . . . . . . . . . . . . . . . . . . . . . 5-27

vii
 Known Parameter Values for a Regression Model with AR Errors . . . . . . 5-28
Regression Model with AR Errors and t Innovations . . . . . . . . . . . . . . . . 5-29

Create Regression Models with MA Errors . . . . . . . . . . . . . . . . . . . . . . . . 5-31

Default Regression Model with MA Errors . . . . . . . . . . . . . . . . . . . . . . . 5-31
MA Error Model Without an Intercept . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-32
MA Error Model with Nonconsecutive Lags . . . . . . . . . . . . . . . . . . . . . . 5-32
Known Parameter Values for a Regression Model with MA Errors . . . . . . 5-33
Regression Model with MA Errors and t Innovations . . . . . . . . . . . . . . . . 5-34

Create Regression Models with ARMA Errors . . . . . . . . . . . . . . . . . . . . . . 5-36

Default Regression Model with ARMA Errors . . . . . . . . . . . . . . . . . . . . . 5-36
ARMA Error Model Without an Intercept . . . . . . . . . . . . . . . . . . . . . . . . 5-37
ARMA Error Model with Nonconsecutive Lags . . . . . . . . . . . . . . . . . . . . 5-37
Known Parameter Values for a Regression Model with ARMA Errors . . . . 5-38
Regression Model with ARMA Errors and t Innovations . . . . . . . . . . . . . 5-38
Specify Regression Model with ARMA Errors Using Econometric Modeler
App . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-40

Create Regression Models with ARIMA Errors . . . . . . . . . . . . . . . . . . . . . 5-44

Default Regression Model with ARIMA Errors . . . . . . . . . . . . . . . . . . . . . 5-44
ARIMA Error Model Without an Intercept . . . . . . . . . . . . . . . . . . . . . . . . 5-45
ARIMA Error Model with Nonconsecutive Lags . . . . . . . . . . . . . . . . . . . . 5-45
Known Parameter Values for a Regression Model with ARIMA Errors . . . 5-46
Regression Model with ARIMA Errors and t Innovations . . . . . . . . . . . . . 5-47

Create Regression Models with SARIMA Errors . . . . . . . . . . . . . . . . . . . . 5-49

SARMA Error Model Without an Intercept . . . . . . . . . . . . . . . . . . . . . . . 5-49
Known Parameter Values for a Regression Model with SARIMA Errors . . 5-50
Regression Model with SARIMA Errors and t Innovations . . . . . . . . . . . . 5-50

Specify Regression Model with SARIMA Errors . . . . . . . . . . . . . . . . . . . . 5-53

Specify ARIMA Error Model Innovation Distribution . . . . . . . . . . . . . . . . 5-60

About the Innovation Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-60
Innovation Distribution Options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-61
Specify Innovation Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-61

Impulse Response of Regression Models with ARIMA Errors . . . . . . . . . 5-65

Plot Impulse Response of Regression Model with ARIMA Errors . . . . . . 5-66

Regression Model with AR Errors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-66
Regression Model with MA Errors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-67
Regression Model with ARMA Errors . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-68
Regression Model with ARIMA Errors . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-70

Maximum Likelihood Estimation of regARIMA Models . . . . . . . . . . . . . . 5-73

Innovation Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-73
Loglikelihood Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-73

regARIMA Model Estimation Using Equality Constraints . . . . . . . . . . . . 5-75

Presample Values for regARIMA Model Estimation . . . . . . . . . . . . . . . . 5-109

Initial Values for regARIMA Model Estimation . . . . . . . . . . . . . . . . . . . . 5-111

viii Contents
Optimization Settings for regARIMA Model Estimation . . . . . . . . . . . . 5-113
Optimization Options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-113
Constraints on Regression Models with ARIMA Errors . . . . . . . . . . . . . 5-115

Estimate a Regression Model with ARIMA Errors . . . . . . . . . . . . . . . . . 5-116

Estimate a Regression Model with Multiplicative ARIMA Errors . . . . . 5-123

Select Regression Model with ARIMA Errors . . . . . . . . . . . . . . . . . . . . . 5-131

Choose Lags for ARMA Error Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-133

Intercept Identifiability in Regression Models with ARIMA Errors . . . 5-137

Intercept Identifiability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-137
Intercept Identifiability Illustration . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-138

Alternative ARIMA Model Representations . . . . . . . . . . . . . . . . . . . . . . . 5-141

regARIMA to ARIMAX Model Conversion . . . . . . . . . . . . . . . . . . . . . . . 5-141
Illustrate regARIMA to ARIMAX Model Conversion . . . . . . . . . . . . . . . . 5-142

Simulate Regression Models with ARMA Errors . . . . . . . . . . . . . . . . . . . 5-147

Simulate an AR Error Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-147
Simulate an MA Error Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-153
Simulate an ARMA Error Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-159

Simulate Regression Models with Nonstationary Errors . . . . . . . . . . . . 5-166

Simulate a Regression Model with Nonstationary Errors . . . . . . . . . . . 5-166
Simulate a Regression Model with Nonstationary Exponential Errors . . 5-169

Simulate Regression Models with Multiplicative Seasonal Errors . . . . 5-174

Simulate a Regression Model with Stationary Multiplicative Seasonal Errors
.................................................... 5-174
Untitled . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-176

Monte Carlo Simulation of Regression Models with ARIMA Errors . . . 5-179

What Is Monte Carlo Simulation? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-179
Generate Monte Carlo Sample Paths . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-179
Monte Carlo Error . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-180

Presample Data for regARIMA Model Simulation . . . . . . . . . . . . . . . . . 5-182

Transient Effects in regARIMA Model Simulations . . . . . . . . . . . . . . . . 5-183

What Are Transient Effects? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-183
Illustration of Transient Effects on Regression . . . . . . . . . . . . . . . . . . . 5-183

Forecast a Regression Model with ARIMA Errors . . . . . . . . . . . . . . . . . . 5-191

Forecast a Regression Model with Multiplicative Seasonal ARIMA Errors

........................................................ 5-194

Verify Predictive Ability Robustness of a regARIMA Model . . . . . . . . . . 5-198

MMSE Forecasting Regression Models with ARIMA Errors . . . . . . . . . 5-200

What Are MMSE Forecasts? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-200
How forecast Generates MMSE Forecasts . . . . . . . . . . . . . . . . . . . . . . 5-200

ix
 Forecast Error . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-202

Monte Carlo Forecasting of regARIMA Models . . . . . . . . . . . . . . . . . . . . 5-203

Monte Carlo Forecasts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-203
Advantage of Monte Carlo Forecasts . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-203

Bayesian Linear Regression

6
Bayesian Linear Regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-2
Classical Versus Bayesian Analyses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-2
Main Bayesian Analysis Components . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-3
Posterior Estimation and Inference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-4

Implement Bayesian Linear Regression . . . . . . . . . . . . . . . . . . . . . . . . . . 6-10

Workflow for Standard Bayesian Linear Regression Models . . . . . . . . . . 6-10
Workflow for Bayesian Predictor Selection . . . . . . . . . . . . . . . . . . . . . . . 6-13

Specify Gradient for HMC Sampler . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-18

Posterior Estimation and Simulation Diagnostics . . . . . . . . . . . . . . . . . . 6-27

Diagnose MCMC Samples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-27
Perform Sensitivity Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-34

Tune Slice Sampler For Posterior Estimation . . . . . . . . . . . . . . . . . . . . . . 6-36

Compare Robust Regression Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . 6-43

Bayesian Lasso Regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-52

Bayesian Stochastic Search Variable Selection . . . . . . . . . . . . . . . . . . . . 6-63

Replacing Removed Syntaxes of estimate . . . . . . . . . . . . . . . . . . . . . . . . . 6-73

Replace Removed Syntax When Estimating Analytical Marginal Posterior
..................................................... 6-74
Replace Removed Syntax When Estimating Numerical Marginal Posterior
..................................................... 6-75
Replace Removed Syntax When Estimating Conditional Posterior . . . . . . 6-77

Conditional Mean Models

7
Conditional Mean Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-3
Unconditional vs. Conditional Mean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-3
Static vs. Dynamic Conditional Mean Models . . . . . . . . . . . . . . . . . . . . . . 7-3
Conditional Mean Models for Stationary Processes . . . . . . . . . . . . . . . . . . 7-3

Specify Conditional Mean Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-5

Default ARIMA Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-5

x Contents
 Specify Nonseasonal Models Using Name-Value Pairs . . . . . . . . . . . . . . . . 7-7
Specify Multiplicative Models Using Name-Value Pairs . . . . . . . . . . . . . . 7-11
Specify Conditional Mean Model Using Econometric Modeler App . . . . . 7-14

Autoregressive Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-17

AR(p) Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-17
Stationarity of the AR Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-17

AR Model Specifications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-19

Default AR Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-19
AR Model with No Constant Term . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-19
AR Model with Nonconsecutive Lags . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-20
ARMA Model with Known Parameter Values . . . . . . . . . . . . . . . . . . . . . . 7-21
AR Model with t Innovation Distribution . . . . . . . . . . . . . . . . . . . . . . . . . 7-22
Specify AR Model Using Econometric Modeler App . . . . . . . . . . . . . . . . 7-22

Moving Average Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-26

MA(q) Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-26
Invertibility of the MA Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-26

MA Model Specifications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-28

Default MA Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-28
MA Model with No Constant Term . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-28
MA Model with Nonconsecutive Lags . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-29
MA Model with Known Parameter Values . . . . . . . . . . . . . . . . . . . . . . . . 7-30
MA Model with t Innovation Distribution . . . . . . . . . . . . . . . . . . . . . . . . 7-31
Specify MA Model Using Econometric Modeler App . . . . . . . . . . . . . . . . 7-31

Autoregressive Moving Average Model . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-35

ARMA(p,q) Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-35
Stationarity and Invertibility of the ARMA Model . . . . . . . . . . . . . . . . . . 7-35

ARMA Model Specifications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-37

Default ARMA Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-37
ARMA Model with No Constant Term . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-37
ARMA Model with Known Parameter Values . . . . . . . . . . . . . . . . . . . . . . 7-38
Specify ARMA Model Using Econometric Modeler App . . . . . . . . . . . . . . 7-39

ARIMA Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-42

ARIMA Model Specifications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-44

Default ARIMA Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-44
ARIMA Model with Known Parameter Values . . . . . . . . . . . . . . . . . . . . . 7-45
Specify ARIMA Model Using Econometric Modeler App . . . . . . . . . . . . . 7-45

Multiplicative ARIMA Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-49

Multiplicative ARIMA Model Specifications . . . . . . . . . . . . . . . . . . . . . . . 7-51

Seasonal ARIMA Model with No Constant Term . . . . . . . . . . . . . . . . . . . 7-51
Seasonal ARIMA Model with Known Parameter Values . . . . . . . . . . . . . . 7-52
Specify Multiplicative ARIMA Model Using Econometric Modeler App . . 7-53

Specify Multiplicative ARIMA Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-58

xi
 ARIMA Model Including Exogenous Covariates . . . . . . . . . . . . . . . . . . . . 7-62
ARIMAX(p,D,q) Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-62
Conventions and Extensions of the ARIMAX Model . . . . . . . . . . . . . . . . . 7-62

ARIMAX Model Specifications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-64

Create ARIMAX Model Using Name-Value Pairs . . . . . . . . . . . . . . . . . . . 7-64
Specify ARMAX Model Using Dot Notation . . . . . . . . . . . . . . . . . . . . . . . 7-65
Specify ARIMAX or SARIMAX Model Using Econometric Modeler App . . 7-66

Modify Properties of Conditional Mean Model Objects . . . . . . . . . . . . . . 7-70

Dot Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-70
Nonmodifiable Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-73

Specify Conditional Mean Model Innovation Distribution . . . . . . . . . . . . 7-75

About the Innovation Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-75
Choices for the Variance Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-75
Choices for the Innovation Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . 7-75
Specify the Innovation Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-76
Modify the Innovation Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-78

Specify Conditional Mean and Variance Models . . . . . . . . . . . . . . . . . . . . 7-80

Impulse Response Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-84

Plot the Impulse Response Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-85

Moving Average Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-85
Autoregressive Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-86
ARMA Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-87

Box-Jenkins Differencing vs. ARIMA Estimation . . . . . . . . . . . . . . . . . . . 7-89

Maximum Likelihood Estimation for Conditional Mean Models . . . . . . . 7-92

Innovation Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-92
Loglikelihood Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-92

Conditional Mean Model Estimation with Equality Constraints . . . . . . . 7-94

Presample Data for Conditional Mean Model Estimation . . . . . . . . . . . . 7-95

Initial Values for Conditional Mean Model Estimation . . . . . . . . . . . . . . 7-97

Optimization Settings for Conditional Mean Model Estimation . . . . . . . 7-99

Optimization Options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-99
Conditional Mean Model Constraints . . . . . . . . . . . . . . . . . . . . . . . . . . 7-101

Estimate Multiplicative ARIMA Model . . . . . . . . . . . . . . . . . . . . . . . . . . 7-102

Model Seasonal Lag Effects Using Indicator Variables . . . . . . . . . . . . . 7-105

Forecast IGD Rate from ARX Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-109

Estimate Conditional Mean and Variance Model . . . . . . . . . . . . . . . . . . 7-115

Choose ARMA Lags Using BIC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-120

xii Contents
 Infer Residuals for Diagnostic Checking . . . . . . . . . . . . . . . . . . . . . . . . . 7-123

Monte Carlo Simulation of Conditional Mean Models . . . . . . . . . . . . . . 7-128

What Is Monte Carlo Simulation? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-128
Generate Monte Carlo Sample Paths . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-128
Monte Carlo Error . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-129

Presample Data for Conditional Mean Model Simulation . . . . . . . . . . . 7-130

Transient Effects in Conditional Mean Model Simulations . . . . . . . . . . 7-131

Simulate Stationary Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-132

Simulate AR Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-132
Simulate MA Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-136

Simulate Trend-Stationary and Difference-Stationary Processes . . . . . 7-140

Simulate Multiplicative ARIMA Models . . . . . . . . . . . . . . . . . . . . . . . . . 7-144

Simulate Conditional Mean and Variance Models . . . . . . . . . . . . . . . . . 7-147

Monte Carlo Forecasting of Conditional Mean Models . . . . . . . . . . . . . 7-151

Monte Carlo Forecasts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-151
Advantage of Monte Carlo Forecasting . . . . . . . . . . . . . . . . . . . . . . . . . 7-151

MMSE Forecasting of Conditional Mean Models . . . . . . . . . . . . . . . . . . 7-152

What Are MMSE Forecasts? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-152
How forecast Generates MMSE Forecasts . . . . . . . . . . . . . . . . . . . . . . 7-152
Forecast Error . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-153

Convergence of AR Forecasts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-155

Forecast Multiplicative ARIMA Model . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-159

Specify Presample and Forecast Period Data To Forecast ARIMAX Model

........................................................ 7-162

Forecast Conditional Mean and Variance Model . . . . . . . . . . . . . . . . . . . 7-166

Model and Simulate Electricity Spot Prices Using the Skew-Normal

Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-169

Conditional Variance Models

8
Conditional Variance Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-2
General Conditional Variance Model Definition . . . . . . . . . . . . . . . . . . . . . 8-2
GARCH Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-3
EGARCH Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-3
GJR Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-4

xiii
 Specify GARCH Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-6
Default GARCH Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-6
Specify Default GARCH Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-7
Using Name-Value Pair Arguments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-8
Specify GARCH Model Using Econometric Modeler App . . . . . . . . . . . . . 8-11
Specify GARCH Model with Mean Offset . . . . . . . . . . . . . . . . . . . . . . . . . 8-13
Specify GARCH Model with Known Parameter Values . . . . . . . . . . . . . . . 8-14
Specify GARCH Model with t Innovation Distribution . . . . . . . . . . . . . . . 8-14
Specify GARCH Model with Nonconsecutive Lags . . . . . . . . . . . . . . . . . . 8-15

Specify EGARCH Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-17

Default EGARCH Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-17
Specify Default EGARCH Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-19
Using Name-Value Pair Arguments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-19
Specify EGARCH Model Using Econometric Modeler App . . . . . . . . . . . . 8-22
Specify EGARCH Model with Mean Offset . . . . . . . . . . . . . . . . . . . . . . . . 8-24
Specify EGARCH Model with Nonconsecutive Lags . . . . . . . . . . . . . . . . . 8-24
Specify EGARCH Model with Known Parameter Values . . . . . . . . . . . . . . 8-25
Specify EGARCH Model with t Innovation Distribution . . . . . . . . . . . . . . 8-26

Specify GJR Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-28

Default GJR Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-28
Specify Default GJR Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-29
Using Name-Value Pair Arguments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-30
Specify GJR Model Using Econometric Modeler App . . . . . . . . . . . . . . . . 8-33
Specify GJR Model with Mean Offset . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-35
Specify GJR Model with Nonconsecutive Lags . . . . . . . . . . . . . . . . . . . . . 8-35
Specify GJR Model with Known Parameter Values . . . . . . . . . . . . . . . . . . 8-36
Specify GJR Model with t Innovation Distribution . . . . . . . . . . . . . . . . . . 8-37

Modify Properties of Conditional Variance Models . . . . . . . . . . . . . . . . . 8-39

Dot Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-39
Nonmodifiable Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-41

Specify the Conditional Variance Model Innovation Distribution . . . . . . 8-44

Specify Conditional Variance Model For Exchange Rates . . . . . . . . . . . . 8-47

Maximum Likelihood Estimation for Conditional Variance Models . . . . 8-52

Innovation Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-52
Loglikelihood Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-52

Conditional Variance Model Estimation with Equality Constraints . . . . 8-54

Presample Data for Conditional Variance Model Estimation . . . . . . . . . . 8-55

Initial Values for Conditional Variance Model Estimation . . . . . . . . . . . . 8-57

Optimization Settings for Conditional Variance Model Estimation . . . . 8-58

Optimization Options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-58
Conditional Variance Model Constraints . . . . . . . . . . . . . . . . . . . . . . . . . 8-60

Infer Conditional Variances and Residuals . . . . . . . . . . . . . . . . . . . . . . . . 8-62

Likelihood Ratio Test for Conditional Variance Models . . . . . . . . . . . . . . 8-66

xiv Contents
 Compare Conditional Variance Models Using Information Criteria . . . . 8-69

Monte Carlo Simulation of Conditional Variance Models . . . . . . . . . . . . 8-72

What Is Monte Carlo Simulation? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-72
Generate Monte Carlo Sample Paths . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-72
Monte Carlo Error . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-73

Presample Data for Conditional Variance Model Simulation . . . . . . . . . . 8-75

Simulate GARCH Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-76

Assess EGARCH Forecast Bias Using Simulations . . . . . . . . . . . . . . . . . . 8-81

Simulate Conditional Variance Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-86

Monte Carlo Forecasting of Conditional Variance Models . . . . . . . . . . . . 8-89

Monte Carlo Forecasts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-89
Advantage of Monte Carlo Forecasting . . . . . . . . . . . . . . . . . . . . . . . . . . 8-89

MMSE Forecasting of Conditional Variance Models . . . . . . . . . . . . . . . . . 8-90

What Are MMSE Forecasts? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-90
EGARCH MMSE Forecasts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-90
How forecast Generates MMSE Forecasts . . . . . . . . . . . . . . . . . . . . . . . 8-90

Forecast GJR Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-94

Forecast a Conditional Variance Model . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-97

Converting from GARCH Functions to Model Objects . . . . . . . . . . . . . . . 8-99

Multivariate Time Series Models

9
Vector Autoregression (VAR) Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-3
Types of Stationary Multivariate Time Series Models . . . . . . . . . . . . . . . . 9-3
Lag Operator Representation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-5
Stable and Invertible Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-6
Models with Regression Component . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-6
VAR Model Workflow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-7

Multivariate Time Series Data Formats . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-10

Multivariate Time Series Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-10
Load Multivariate Economic Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-10
Multivariate Data Format . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-12
Preprocess Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-14
Time Base Partitions for Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-15
Partition Multivariate Time Series Data for Estimation . . . . . . . . . . . . . . 9-17

Vector Autoregression (VAR) Model Creation . . . . . . . . . . . . . . . . . . . . . . 9-19

Create VAR Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-19
Fully Specified Model Object . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-20
Model Template for Unrestricted Estimation . . . . . . . . . . . . . . . . . . . . . . 9-22

xv
 Partially Specified Model Object for Restricted Estimation . . . . . . . . . . . 9-23
Display and Change Model Objects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-23
Select Appropriate Lag Order . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-26

Create and Adjust VAR Model Using Shorthand Syntax . . . . . . . . . . . . . . 9-29

Create and Adjust VAR Model Using Longhand Syntax . . . . . . . . . . . . . . 9-31

VAR Model Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-33

Preparing VAR Models for Fitting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-33
Fitting Models to Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-33
Examining the Stability of a Fitted Model . . . . . . . . . . . . . . . . . . . . . . . . 9-34

Convert VARMA Model to VAR Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-36

Fit VAR Model of CPI and Unemployment Rate . . . . . . . . . . . . . . . . . . . . 9-37

Fit VAR Model to Simulated Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-41

VAR Model Forecasting, Simulation, and Analysis . . . . . . . . . . . . . . . . . . 9-43

VAR Model Forecasting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-43
Data Scaling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-45
Calculating Impulse Responses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-45

Generate VAR Model Impulse Responses . . . . . . . . . . . . . . . . . . . . . . . . . 9-47

Compare Generalized and Orthogonalized Impulse Response Functions

......................................................... 9-51

Forecast VAR Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-59

Forecast VAR Model Using Monte Carlo Simulation . . . . . . . . . . . . . . . . 9-62

Forecast VAR Model Conditional Responses . . . . . . . . . . . . . . . . . . . . . . . 9-65

Implement Seemingly Unrelated Regression . . . . . . . . . . . . . . . . . . . . . . 9-69

Estimate Capital Asset Pricing Model Using SUR . . . . . . . . . . . . . . . . . . 9-74

Simulate Responses of Estimated VARX Model . . . . . . . . . . . . . . . . . . . . 9-77

Simulate VAR Model Conditional Responses . . . . . . . . . . . . . . . . . . . . . . 9-84

Simulate Responses Using filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-88

VAR Model Case Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-90

Convert from vgx Functions to Model Objects . . . . . . . . . . . . . . . . . . . . 9-104

Cointegration and Error Correction Analysis . . . . . . . . . . . . . . . . . . . . . 9-107

Integration and Cointegration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-107
Cointegration and Error Correction . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-107
The Role of Deterministic Terms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-108
Cointegration Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-109

xvi Contents
 Determine Cointegration Rank of VEC Model . . . . . . . . . . . . . . . . . . . . 9-111

Identifying Single Cointegrating Relations . . . . . . . . . . . . . . . . . . . . . . . 9-113

The Engle-Granger Test for Cointegration . . . . . . . . . . . . . . . . . . . . . . . 9-113
Limitations of the Engle-Granger Test . . . . . . . . . . . . . . . . . . . . . . . . . . 9-113

Test for Cointegration Using the Engle-Granger Test . . . . . . . . . . . . . . 9-116

Estimate VEC Model Parameters Using egcitest . . . . . . . . . . . . . . . . . . 9-120

VEC Model Monte Carlo Forecasts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-123

Generate VEC Model Impulse Responses . . . . . . . . . . . . . . . . . . . . . . . . 9-131

Identifying Multiple Cointegrating Relations . . . . . . . . . . . . . . . . . . . . . 9-135

Test for Cointegration Using the Johansen Test . . . . . . . . . . . . . . . . . . . 9-136

Estimate VEC Model Parameters Using jcitest . . . . . . . . . . . . . . . . . . . . 9-138

Compare Approaches to Cointegration Analysis . . . . . . . . . . . . . . . . . . . 9-141

Testing Cointegrating Vectors and Adjustment Speeds . . . . . . . . . . . . . 9-144

Test Cointegrating Vectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-145

Test Adjustment Speeds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-147

Structural Change Models

10
Discrete-Time Markov Chains . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-2
What Are Discrete-Time Markov Chains? . . . . . . . . . . . . . . . . . . . . . . . . 10-2
Discrete-Time Markov Chain Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-3

Markov Chain Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-8

Discrete-Time Markov Chain Object Framework Overview . . . . . . . . . . . 10-8
Markov Chain Analysis Workflow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-11

Create and Modify Markov Chain Model Objects . . . . . . . . . . . . . . . . . . 10-17

Create Markov Chain from Stochastic Transition Matrix . . . . . . . . . . . . 10-17
Create Markov Chain from Random Transition Matrix . . . . . . . . . . . . . 10-19
Specify Structure for Random Markov Chain . . . . . . . . . . . . . . . . . . . . 10-20

Work with State Transitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-23

Visualize Markov Chain Structure and Evolution . . . . . . . . . . . . . . . . . . 10-27

Determine Asymptotic Behavior of Markov Chain . . . . . . . . . . . . . . . . . 10-39

Identify Classes in Markov Chain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-47

xvii
 Compare Markov Chain Mixing Times . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-50

Simulate Random Walks Through Markov Chain . . . . . . . . . . . . . . . . . . 10-59

Compute State Distribution of Markov Chain at Each Time Step . . . . . 10-66

State-Space Models
11
What Are State-Space Models? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-3
Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-3
State-Space Model Creation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-5

What Is the Kalman Filter? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-7

Standard Kalman Filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-7
State Forecasts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-8
Filtered States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-8
Smoothed States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-9
Smoothed State Disturbances . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-9
Forecasted Observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-10
Smoothed Observation Innovations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-10
Kalman Gain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-11
Backward Recursion of the Kalman Filter . . . . . . . . . . . . . . . . . . . . . . . 11-11
Diffuse Kalman Filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-12

Explicitly Create State-Space Model Containing Known Parameter Values

........................................................ 11-13

Create State-Space Model with Unknown Parameters . . . . . . . . . . . . . . 11-15

Explicitly Create State-Space Model Containing Unknown Parameters
.................................................... 11-15
Implicitly Create Time-Invariant State-Space Model . . . . . . . . . . . . . . . 11-16

Create State-Space Model Containing ARMA State . . . . . . . . . . . . . . . . 11-18

Implicitly Create State-Space Model Containing Regression Component

........................................................ 11-21

Implicitly Create Diffuse State-Space Model Containing Regression

Component . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-23

Implicitly Create Time-Varying State-Space Model . . . . . . . . . . . . . . . . 11-25

Implicitly Create Time-Varying Diffuse State-Space Model . . . . . . . . . . 11-27

Create State-Space Model with Random State Coefficient . . . . . . . . . . 11-30

Estimate Time-Invariant State-Space Model . . . . . . . . . . . . . . . . . . . . . . 11-32

Estimate Time-Varying State-Space Model . . . . . . . . . . . . . . . . . . . . . . . 11-35

xviii Contents
Estimate Time-Varying Diffuse State-Space Model . . . . . . . . . . . . . . . . . 11-39

Estimate State-Space Model Containing Regression Component . . . . . 11-43

Filter States of State-Space Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-45

Filter Time-Varying State-Space Model . . . . . . . . . . . . . . . . . . . . . . . . . . 11-48

Filter Time-Varying Diffuse State-Space Model . . . . . . . . . . . . . . . . . . . 11-53

Filter States of State-Space Model Containing Regression Component

........................................................ 11-59

Smooth States of State-Space Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-62

Smooth Time-Varying State-Space Model . . . . . . . . . . . . . . . . . . . . . . . . 11-65

Smooth Time-Varying Diffuse State-Space Model . . . . . . . . . . . . . . . . . . 11-71

Smooth States of State-Space Model Containing Regression Component

........................................................ 11-77

Simulate States and Observations of Time-Invariant State-Space Model

........................................................ 11-80

Simulate Time-Varying State-Space Model . . . . . . . . . . . . . . . . . . . . . . . 11-83

Simulate States of Time-Varying State-Space Model Using Simulation

Smoother . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-87

Estimate Random Parameter of State-Space Model . . . . . . . . . . . . . . . . 11-90

Forecast State-Space Model Using Monte-Carlo Methods . . . . . . . . . . . 11-97

Forecast State-Space Model Observations . . . . . . . . . . . . . . . . . . . . . . 11-103

Forecast Observations of State-Space Model Containing Regression

Component . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-106

Forecast Time-Varying State-Space Model . . . . . . . . . . . . . . . . . . . . . . 11-110

Forecast State-Space Model Containing Regime Change in the Forecast

Horizon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-114

Forecast Time-Varying Diffuse State-Space Model . . . . . . . . . . . . . . . . 11-119

Compare Simulation Smoother to Smoothed States . . . . . . . . . . . . . . 11-123

Rolling-Window Analysis of Time-Series Models . . . . . . . . . . . . . . . . . 11-128

Rolling-Window Analysis for Parameter Stability . . . . . . . . . . . . . . . . . 11-128
Rolling Window Analysis for Predictive Performance . . . . . . . . . . . . . . 11-128

Assess State-Space Model Stability Using Rolling Window Analysis . 11-131

Assess Model Stability Using Rolling Window Analysis . . . . . . . . . . . . 11-131

xix
 Assess Stability of Implicitly Created State-Space Model . . . . . . . . . . 11-134

Choose State-Space Model Specification Using Backtesting . . . . . . . . 11-138

Functions
12

Appendices
A
Data Sets and Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-2

Glossary

xx Contents
 1

Getting Started

• “Econometrics Toolbox Product Description” on page 1-2

• “Econometric Modeling” on page 1-3
• “Econometrics Toolbox Model Objects, Properties, and Object Functions” on page 1-7
• “Stochastic Process Characteristics” on page 1-16
• “Bibliography” on page 1-22
1 Getting Started

Econometrics Toolbox Product Description

Model and analyze financial and economic systems using statistical methods

Econometrics Toolbox provides functions for modeling and analyzing time series data. It offers a wide
range of diagnostic tests for model selection, including tests for impulse analysis, unit roots and
stationarity, cointegration, and structural change. You can estimate, simulate, and forecast economic
systems using a variety of models, including, regression, ARIMA, state space, GARCH, multivariate
VAR and VEC, and switching models representing dynamic shifts in data. The toolbox also provides
Bayesian and Markov-based tools for developing time-varying models that learn from new data.

1-2
 Econometric Modeling

Econometric Modeling
In this section...
“Model Selection” on page 1-3
“Econometrics Toolbox Features” on page 1-3

Model Selection
A probabilistic time series model is necessary for a wide variety of analysis goals, including
regression inference, forecasting, and Monte Carlo simulation. When selecting a model, aim to find
the most parsimonious model that adequately describes your data. A simple model is easier to
estimate, forecast, and interpret.

• Specification tests help you identify one or more model families that could plausibly describe the
data generating process.
• Model comparisons help you compare the fit of competing models, with penalties for complexity.
• Goodness-of-fit checks help you assess the in-sample adequacy of your model, verify that all model
assumptions hold, and evaluate out-of-sample forecast performance.

Model selection is an iterative process. When goodness-of-fit checks suggest model assumptions are
not satisfied—or the predictive performance of the model is not satisfactory—consider making model
adjustments. Additional specification tests, model comparisons, and goodness-of-fit checks help guide
this process.

Econometrics Toolbox Features

Modeling Features Related Functions
Questions
What is the • The conditional mean and variance models, regression • arima
dimension of my models with ARIMA errors, and Bayesian linear regression • bayeslm
response variable? models in this toolbox are for modeling univariate,
discrete-time data. • egarch

• Separate models are available for multivariate, discrete- • egcitest

time data, such as VAR and VEC models. • dssm
• State-space models support univariate or multivariate • garch
response variables. • gjr
• jcontest
• regARIMA
• ssm
• varm
Is my time series • Stationarity tests are available. If your data is not • arima
stationary? stationary, consider transforming your data. Stationarity is • i10test
the foundation of many time series models.
• kpsstest
• Or, consider using a nonstationary ARIMA model if there
is evidence of a unit root in your data. • lmctest

1-3
1 Getting Started

Modeling Features Related Functions

Questions
Does my time series • Unit root tests are available. Evidence in favor of a unit • adftest
have a unit root? root suggests your data is difference stationary. • arima
• You can difference a series with a unit root until it is • i10test
stationary, or model it using a nonstationary ARIMA
model. • pptest
• vratiotest
How can I handle • You can deseasonalize (seasonally adjust) your data. Use • arima
seasonal effects? seasonal filters or regression models to estimate the • regARIMA
seasonal component.
• Seasonal ARIMA models use seasonal differencing to
remove seasonal effects. You can also include seasonal
lags to model seasonal autocorrelation (both additively
and multiplicatively).
Is my data • Sample autocorrelation and partial autocorrelation • arima
autocorrelated? functions help identify autocorrelation. • autocorr
• Conduct a Ljung-Box Q-test to test autocorrelations at • fgls
several lags jointly.
• hac
• If autocorrelation is present, consider using a conditional
mean model. • lbqtest

• For regression models with autocorrelated errors, • parcorr

consider using FGLS or HAC estimators. If the error • regARIMA
model structure is an ARIMA model, consider using a
regression model with ARIMA errors.
What if my data is • Looking for autocorrelation in the squared residual series • archtest
heteroscedastic is one way to detect conditional heteroscedasticity. • egarch
(exhibits volatility • Engle’s ARCH test evaluates evidence against the null of • fgls
clustering)? independent innovations in favor of an ARCH model
alternative. • garch

• To model conditional heteroscedasticity, consider using a • gjr

conditional variance model. • hac
• For regression models that exhibit heteroscedastic errors,
consider using FGLS or HAC estimators.
Is there an • You can use a Student’s t distribution to model fatter tails • arima
alternative to a than a Gaussian distribution (excess kurtosis). • egarch
Gaussian innovation • You can specify a t innovation distribution for all
distribution for • garch
conditional mean and variance models, and ARIMA error
leptokurtic data? models in Econometrics Toolbox. • gjr

• You can estimate the degrees of freedom of the t • regARIMA

distribution along with other model parameters.

1-4
 Econometric Modeling

Modeling Features Related Functions

Questions
How do I decide • You can compare nested models using misspecification • aicbic
between several tests, such as the likelihood ratio test, Wald’s test, or • lmtest
model fits? Lagrange multiplier test.
• lratiotest
• Information criteria, such as AIC or BIC, compare model
fit with a penalty for complexity. • waldtest
Do I have two or • The Johansen and Engle-Granger cointegration tests • egcitest
more time series that assess evidence of cointegration. • jcitest
are cointegrated? • Consider using the VEC model for modeling multivariate, • jcontest
cointegrated series.
• Also consider cointegration when regressing time series.
If present, it can introduce spurious regression effects.
What if I want to • ARIMAX, VARX, regression models with ARIMA errors, • arima
include predictor and Bayesian linear regression models are available in • bayeslm
variables? this toolbox.
• dssm
• State-space models support predictor data.
• ssm
• regARIMA
• varm
What if I want to • Regression models with ARIMA errors are available in this • bayeslm
implement toolbox. • fgls
regression, but the • Regress robustly using FGLS or HAC estimators.
classical linear • hac
model assumptions • Use Bayesian linear regression. • mvregress
might not apply? • For a series of examples on time series regression • regARIMA
techniques that illustrate common principles and tasks in
time series regression modeling, see Econometrics
Toolbox Examples.
• For more regression options, see Statistics and Machine
Learning Toolbox™ documentation.
What if observations Standard, linear state-space modeling is available in this • dssm
of a dynamic process toolbox. • ssm
include
measurement error?

See Also

Related Examples
• “Box-Jenkins Model Selection” on page 3-4
• “Detect Autocorrelation” on page 3-15
• “Detect ARCH Effects” on page 3-22
• “Unit Root Tests” on page 3-32
• “Time Series Regression I: Linear Models”

1-5
1 Getting Started

• “Time Series Regression II: Collinearity and Estimator Variance”

• “Time Series Regression III: Influential Observations”
• “Time Series Regression IV: Spurious Regression”
• “Time Series Regression V: Predictor Selection”
• “Time Series Regression VI: Residual Diagnostics”
• “Time Series Regression VII: Forecasting”
• “Time Series Regression VIII: Lagged Variables and Estimator Bias”
• “Time Series Regression IX: Lag Order Selection”
• “Time Series Regression X: Generalized Least Squares and HAC Estimators”

More About
• “Trend-Stationary vs. Difference-Stationary Processes” on page 2-6
• “Box-Jenkins Methodology” on page 3-2
• “Goodness of Fit” on page 3-63
• “Regression Models with Time Series Errors” on page 5-5
• “Nonspherical Models” on page 3-67
• “Conditional Mean Models” on page 7-3
• “Conditional Variance Models” on page 8-2
• “Vector Autoregression (VAR) Models” on page 9-3
• “Cointegration and Error Correction Analysis” on page 9-107

1-6
 Econometrics Toolbox Model Objects, Properties, and Object Functions

Econometrics Toolbox Model Objects, Properties, and Object

Functions
In this section...
“Model Objects” on page 1-7
“Model Properties” on page 1-8
“Specify Models” on page 1-10
“Retrieve Model Properties” on page 1-13
“Modify Model Properties” on page 1-14
“Object Functions” on page 1-15

Model Objects
After you have a potential model for your data, you must specify the model to MATLAB® to proceed
with your analysis. Econometrics Toolbox has model objects for storing discrete-time econometric
models.

For univariate series, the available model objects are:

• arima — for integrated, autoregressive, moving average (ARIMA) models optionally containing
exogenous predictor variables
• garch — for generalized autoregressive conditional heteroscedasticity models (GARCH)
• egarch — for exponential GARCH models
• gjr — for Glosten-Jagannathan-Runkle models
• regARIMA — for regression models with ARIMA errors

For multivariate series, the available model objects are:

• varm — for vector autoregression models optionally containing exogenous predictor variables
• vecm — for vector error-correction (cointegrated VARM) models optionally containing exogenous
predictor variables

Econometrics Toolbox supports univariate Bayesian linear regression analysis. Bayesian linear
regression model objects specify the joint prior distribution of the regression coefficients and
disturbance variance. The available prior model objects are:

• conjugateblm — for the normal-inverse-gamma conjugate prior model. The regression

coefficients and disturbance variance are dependent random variables.
• semiconjugateblm — for the normal-inverse-gamma semiconjugate prior model. The regression
coefficients and disturbance variance are independent random variables.
• diffuseblm — the joint prior distribution is proportional to the inverse of the disturbance
variance.
• empiricalblm — the joint prior distribution is specified by a random sample from the joint
posterior distribution.
• customblm — the joint prior distribution is specified in a custom function that you declare.

To perform Bayesian variable selection, available prior model objects are:

1-7
1 Getting Started

• mixconjugateblm — for performing stochastic search variable selection (SSVS). The regression
coefficients and disturbance variance are dependent random variables (the prior and posterior
distributions are conjugate).
• mixsemiconjugateblm — for performing SSVS. The regression coefficients and disturbance
variance are independent random variables (the prior and posterior distributions are
semiconjugate).
• lassoblm — for performing Bayesian lasso regression.

Econometrics Toolbox supports modelling and analyzing discrete or continuous state Markov models.
Available model objects are:

• dtmc — for discrete-time Markov chain models characterized by transition matrices.

• ssm — for continuous, multivariate state-space models optionally containing exogenous predictor
variables
• dssm — for continuous, multivariate state-space models with diffuse initial states optionally
containing exogenous predictor variables

To create a model object, specify the form of your model to one of the model functions (e.g., arima or
garch). The function creates the model object of the corresponding type in the MATLAB workspace,
as shown in the figure.

You can work with model objects as you would with any other variable in MATLAB. For example, you
can assign the object variable a name, view it in the MATLAB Workspace, and display its value in the
Command Window by typing its name.

This image shows a workspace containing an arima model named Mdl.

Model Properties
A model object holds all the information necessary to estimate, simulate, and forecast econometric
models. This information includes the:

• Parametric form of the model

• Number of model parameters (e.g., the degree of the model)

1-8
 Econometrics Toolbox Model Objects, Properties, and Object Functions

• Innovation distribution (Gaussian or Student’s t)

• Amount of presample data needed to initialize the model

Such pieces of information are properties of the model, which are stored as fields within the model
object. In this way, a model object resembles a MATLAB data structure (struct array).

The five model types—arima, garch, egarch, gjr, and regARIMA—have properties according to the
econometric models they support. Each property has a predefined name, which you cannot change.

For example, arima supports conditional mean models (multiplicative and additive AR, MA, ARMA,
ARIMA, and ARIMAX processes). Every arima model object has these properties, shown with their
corresponding names.

Property Name Property Description

Constant Model constant
AR Nonseasonal AR coefficients
MA Nonseasonal MA coefficients
SAR Seasonal AR coefficients (in a multiplicative model)
SMA Seasonal MA coefficients (in a multiplicative model)
D Degree of nonseasonal differencing
Seasonality Degree of seasonal differencing
Variance Variance of the innovation distribution
Distribution Parametric family of the innovation distribution
P Amount of presample data needed to initialize the AR component of
the model
Q Amount of presample data needed to initialize the MA component of
the model

When a model object exists in the workspace, double-click its name in the Workspace window to open
the Variable Editor. The Variable Editor shows all model properties and their names.

1-9
Another Random Scribd Document
with Unrelated Content
articulation with the nasal bone and extending backward as far as
the view from the nare will allow.
Through this opening, the skin overlying the bone is raised by dull
dissection. A fine nasal saw is next introduced through the
submucous wound and several vertical incisions are made into or
even through the bone about three sixteenths of an inch apart,
dividing the convexed osseous tissue into several sections adherent
at their superior extremity which lies inferior to the insertion of the
levator labii superioris alæque nasi muscle.
A forceps, such as Adams’s, is now introduced and each section of
bone thus made is fractured from below upwards inwardly to
produce a concavity of the osseous tissue.
The operation requires considerable dexterity. The amplitude of
the sawing movement is very much restricted, because of the
palpebral muscular attachment just above.
A frail bone cutting forceps may be employed and the lower half of
the process be removed to avoid encroachment upon the middle
meatus, but this is rarely necessary, as that chamber is found
unusually wide in this case. If the bone is removed, the remaining
bone may be cut into sections, as described, or by the cutting
forceps, and fractured backwards as described.
Retention dressings must be resorted to, to keep the fragments of
the bone in their new position until cicatrization has been sufficiently
established to keep them in place.
When possible Roe advises sawing off the convexity submucously
and, after loosening the skin over the dorsum of the nose, to move
the bony plate thus made over to the opposite side of the nose and
into the concavity usually found there in these cases. If there be no
deviation at the latter site the bone plate can be entirely removed
through the inferior wound in the mucosa.
 CHAPTER XVII
ELECTROLYSIS IN DERMATOLOGY

Several references have been made in the preceding chapter to

the specific use of electricity without a description, however, of its
source or application. The author does not deem it necessary in this
volume to go into the principles of electricity, and takes it for granted
that the practitioner is sufficiently familiar with a knowledge of the
rudiments of the subject and that he understands the meaning of an
electric cell commonly known as a battery.
The Electric Battery.—An electric cell or battery is made up of
two poles which are named positive, designated by the + (plus)
sign, and negative by the - (minus) sign. In the usual form of cell
used the parts are made up of a carbon and zinc cylinder placed into
a glass jar containing the electrolyte or actuating fluid. The latter is
either an aqueous solution of potassium bichromate or salammoniac
contained in a glass jar.
For continuous use or open circuit work the Le Clanche type of cell
is most practicable.
 Fig. 502.—Electric Wet Cell.

In Fig. 502 a cell of this type is shown in which the positive pole
or element is composed of a solid piece of carbon forming a cover to
the glass jar as well, and the negative element is of zinc. The
covering over of the jar prevents evaporation of the solution and
adds much to its life.
The Voltage or Electromotive Force.—The voltage or
electromotive force from such a cell averages about 1.5 volts.
Voltage represents the force or propelling power of current known
scientifically as the electromotive force and designated EMF. Owing
to the great resistance of the body to the electric current, a
proportionate force is required to attain therapeutic results.
The unit measure of the quantity of current is known as the
ampère. As this is too great for therapeutic use, the thousandth part,
or milliampère, is employed, and for the purpose of measuring the
amount of current given the patient the milliampèremeter is included
in the circuit or flow of current.
 The unit of resistance is termed the Ohm, and to simplify the
method of electrotherapeutic administration the practitioner may
refer to Ohm’s law as a guide. He must remember the average
resistance to the current of the parts to be operated on by this
process. The law is as follows:

EMF or Voltage
C or Current in Ampères = —————————
R or Resistance,

or commonly written

R
C = ———
EMF

The Rheostat.—When we consider that the resistance between

electrodes placed on the palm of the left hand and the back of the
neck is about 4,000 Ohms, it may be readily understood that
considerable voltage is required to overcome this resistance before
the proper amount of current can be employed. Since each cell, for
quick reference, may be said to represent one volt, at least twelve
and not more than sixteen cells would be required for electrolysis.
Not all of the current given off by a battery of such number of cells
should be used on a patient for electrolytic purpose. Some method
must be employed to reduce this voltage and to control it at will.
This is necessary since the life of a cell varies and its current
capacity is limited according to the use the cell is put to. An
instrument of this nature is called a rheostat and is usually made of
graphite or metal wire. Water resistances are also used, but they do
not permit of a constant current because of the consequent heating
and decomposition of the water into its elements at the two metal
poles exposed to the water. The proper instrument will be referred to
later.
 The electric cell represents a certain voltage; to add to this more
cells are needed and connected with each other so that each adds
its voltage to other or the circuit. The method of connecting cells in
this manner is called series connection, in which the carbon element
of one cell is connected with the zinc of the next, and so forth, until
the last cell, leaving two free poles, one carbon and a zinc to which
the wires to hold the electrodes for the patient are connected. As
has been said, the carbon is the positive pole and the zinc the
negative. The method of connection is shown in Fig. 503.

Fig. 503.—Series Connection.

These two poles when brought in contact with human tissue

exhibit different action and effect. Without going into electro-
chemistry it may be said the current of the positive pole is sedative
and that of the negative irritant or destructive. That oxygen and
acids are freed at the positive pole and hydrogen and alkalies at the
negative pole.
It is due to these properties of the current that it is employed
therapeutically, but to properly employ it the current must be
controlled so that the exact amount given or used can be estimated.
This is accomplished first of all by the interposition of resistance
within the circuit. This resistance should be such that the current can
be increased or decreased at will. It has been referred to and is
called a rheostat. Its position in the circuit is shown in Fig. 504.
 Fig. 504.—Shunt Rheostat Connection.

Cell Selector.—The physician may do without such a rheostat

and use a cell selector with the object of adding one or more cells to
the circuit at will. Such instrument is composed of a marble or
wooden base with a number of disks upon it, each disk representing
a cell of the battery. A metal arm is made to slide over these disks,
and as it advances over each disk the current from that cell is added
to the circuit. It may have a second arm which is used to cut out the
current from the cell or cells at the beginning of the circuit—in fact,
will permit of the selection of any cell in the circuit by proper
manipulation. Such a selector is shown in Fig. 505.
 Fig. 505.—Cell Selector.

The connection of the cells of the battery when a selector is used

varies from that just mentioned. The proper wiring with the disks of
the Selector is shown in Fig. 506.
 Fig. 506.—Cell Selector and Battery Arrangement.

Milliampèremeter.—The fact that a proper resistance has been

forced in circuit is not alone sufficient to permit of the proper use of
current for electrolysis. A measuring device should be included, as
has been referred to and called the Milliampèremeter or
Milliammeter. It is shown in Fig. 507.
 Fig. 507.—Milliampèremeter.

The method of connecting this instrument in series with the

current from the rheostat has been shown in Fig. 509.
The Electric Current.—Where the operating room of the
physician is provided with street current it will be found more
economical and cleaner to use that current for this purpose.
Usually the direct current is furnished of a voltage varying from
100 to 125 volts. To utilize such a current a wall plate is employed
and connected to the circuit, as shown in Fig. 508. The resistance of
an electric lamp is added to guard against injuring the patient if by
any accident or negligence the circuit has been improperly closed.
 Fig. 508.—Direct Current Switch Board or Wall
Plate.

Whether the street or battery current is used with such a plate

makes no difference except that with a battery circuit the lamp is not
used. The connections are given in Fig. 509.
 Fig. 509.—Wall-Plate Connections.

It will be observed that a current changing switch has been added

to the wall plate. This is included in the circuit to permit of changing
the poles to the patient without interfering or disconnecting the
electrodes if desired at any time during treatment.
Portable Batteries.—The above instruments and circuits refer to
those to be used in the operating room and are stationary. The
physician may be called upon to treat a patient at a distance and for
this purpose must have a portable battery.
There are many such instruments on the market of both dry and
moist cell type. The moist cells usually require a bichromate of soda
or potash solution and are so constructed that the carbon and zinc
poles are taken out of the electrolyte or solution and placed into
water-tight compartments provided for them. Such an apparatus is
shown in Fig. 510a.

Fig. 510a.—Portable Wet Cell Direct Current Apparatus.

The best cell for this purpose is the silver chloride battery. It is
compact, light in weight, and gives a steady current. The only
objection is the high cost.
Portable batteries should be furnished with a milliampèremeter. A
type of a compact dry cell direct current apparatus is shown in Fig.
510b. In the end the best apparatus proves the most economical.
Fig. 510b.—Direct Current Dry Cell Apparatus with Rheostat and Interrupted Current
Attachment.

Electrodes.—Having the circuit or current under control, it now

becomes necessary to attach electrodes to the free poles to be able
to properly apply it to the patient. These electrodes vary
considerably according to their use. The author will refer to only
those that are of service in electrolysis.
Sponge Electrode.—The one electrode held by the patient is
usually made of a metal disk covered with felt or sponge attached to
a wooden handle and is shown in Fig. 511.
 Fig. 511.—Sponge Electrode.

This electrode represents the positive; the negative pole is held by

the operator. When used, the felt or sponge is moistened with warm
water to which a little salt has been added and is placed into the
palm of the hand, sponge inward.
The author prefers to use a plain metal disk with the sponge and
places a piece of absorbent cotton or gauze over it when in use for
hygienic reason.
When the operator prefers he may resort to arm or wrist
electrodes which can be clamped upon the limb and be held in
position and shown in Fig. 512.

Fig. 512.—Arm Electrode.

The hand electrode is of greater service since the patient can

regulate or make and break the current at will, a matter of no small
consequence when fairly large currents are being used to destroy a
growth upon the skin of the face.
Needles and Needle Holders.—For the negative electrode the
operator uses a needle holder with a needle of proper form and
material.
 Two needle holders are shown in Figs. 513 and 514.

Fig. 513. Fig. 514.

Electrolytic Needle Holders.

When the operator desires he may employ an interrupting needle

holder with which he can make and break the current at will during
the operation. It is shown in Fig. 515.

Fig. 515.—Interrupting Current Needle Holder.

Such a device is not advocated, since the patient is liable to jump

as the current is made suddenly, because of the sharp stinging pain
felt at the point when the needle has entered the tissue or hair
follicle, often resulting in the breaking of the needle and possible
injury to the patient.
Other operators employ a small magnifying glass which may be
attached to the holder, as in Fig. 516, and by a sliding arrangement
be moved up or down the handle to adjust the lens to the proper
focus. This arrangement is indeed novel and may be of service in
removing fine superfluous hairs, but the author has never resorted
to the method.
 Fig. 516.—Needle Holder with Magnifying Glass.

The proper kind of needle to be used for electrolysis varies with

the device of the operator. The ordinary cambric needle usually
advocated is too stiff and thick. Jeweler’s broaches are better, but
are very brittle and easily broken. The ideal needle should be very
thin and made of platinum or irido-platinum. The author prefers the
sharp to the bulbous-pointed. For the removal of other blemishes
than hair from the face the sharp needle only can be used.

REMOVAL OF SUPERFLUOUS HAIR

The moistened sponge electrode connected to the (+) positive
pole of the circuit is placed into the hand of the patient, who lies in a
chair with her head on a level with the physician’s chin when
operating. The light should be southern, or such that the shafts of
the hairs show plainly.
The operator turns on the current, holding the needle holder in
the right hand which is connected by a flexible cord to the (-)
negative pole. The rheostat handle is brought back so that just the
least current is flowing. The needle is now thrust down into the
follicle containing the hair. This must be done very gently so as to
feel when the papilla has been reached by the needle. The depth to
which the needle goes varies very much according to the size and
place of the hair. It may be less than one eighth and more than one
fourth inch.
The patient holding the sponge will at once feel a stinging
sensation when the needle enters the skin, which is later not as
objectionable. The current is now increased by advancing the handle
of the rheostat until about eight milliampères are shown by the
index on the dial.
Within a few seconds a white froth will issue from the follicle,
showing that decomposition of tissue is taking place. The operator
must familiarize himself with the time and amount of current
required to destroy superfluous hairs. Coarse hairs may require as
much as fourteen milliampères, but it is advisable to use a moderate
amount of current and to leave the needle a little longer in the
follicle to avoid scarring of the skin.
The papilla having presumably been destroyed, the patient
loosens her grip on the sponge and the needle is withdrawn.
The operator now takes up an epilating forceps, such as shown in
Fig. 517, and removes the hair. If the hair does not come out of the
follicle readily it shows that it has not been destroyed, and the same
treatment, just described, must be repeated, but for a shorter
duration.

Fig. 517.—Epilating Forceps.

When the hair is removed it will show more or less bulb according
to its size and nourishment.
The physician now proceeds to remove the coarse hairs first. Hairs
should not be removed too closely placed, as the current will destroy
the tissue between the follicle and cause scarring. It is better to
remove the hairs some distance apart, leaving the remaining hairs
for later sittings.
About forty or fifty hairs may be removed at one sitting. This will
require from half to an hour and a half of time, but the operator will
soon accomplish considerable work in a minimum of time.
 Some of the hairs removed will return, showing as black or dark
specks in the skin, in from five to ten days. The number returning
depends on the operator’s skill. At first he should not be surprised to
see fifty per cent come back, but this ratio is reduced so that only
three or four hairs out of fifty may return, and perhaps these stunted
in growth.
The electrolytic removal of hair does not stimulate the growth of
the finer hairs of the skin; that general belief has been erroneous.
Where there is considerable hair to be removed, as with a beard
on a woman’s face, several sittings may be given a week and at
different parts of the face, but with the average patient only one
sitting should be given each week.
More or less edema follows the removal of hair, which may remain
for a day or more. Warm applications will help to remove it.
The operator should at no time state a definite fee to remove the
hair on the face, unless he is certain of the number present. Such
judgment is, indeed, very misleading.

REMOVAL OF MOLES OR OTHER FACIAL

GROWTHS
Moles, warts, fibromata, fungoids, and other excrescences are
best removed with this method, especially where they are of the
nonpedunculated type. It is hardly necessary to state that very light
currents should be used for the light flat growths, such as a dark
freckle or a small yellow mole. The amount of current required varies
from 6 to 24 milliampères, according to the size of the body to be
removed.
The same procedure as with the removal of hairs is followed.
Positive electrode in the hand of the patient, negative pole to the
needle holder. The needle is thrust through the growth on a plane
with the skin and slightly above it. The current will at once produce
a pale color in the mass and white froth will issue about the shaft of
the needle. A comparatively greater amount of current is needed for
this purpose than with the destruction of hairs. The operator must
judge the amount and time required from experience.
The mass is punctured in stellate fashion to assure an even
necrosis, as shown in Fig. 518.

Fig. 518.—Electrolysis Method for Destroying Growths.

The mass will appear much softer after this treatment, is in some
cases, as with flat moles, quite friable, but this disappears in a few
hours and the mass begins to shrivel and dry up, forming a scab,
which is between brown and almost black in color. This scab falls off
in several days, according to its size, leaving a pink eschar, which
gradually turns white and shows very little, if the growth has not
been too large and the electrolysis carefully done. If little
tumefactions, or tips of tissue, still appear, they are removed as soon
after the scab falls off as deemed advisable by the same method.
Warts show more or less recurrence.

TELANGIECTASIS
In this condition there appear in the skin one or many dilated
capillaries. It is quite common about the sides and lobule of the nose
and just inferior to the malar prominence of the cheeks. To destroy
these the fine platinum needle is thrust through the skin and directly
through the canal of the vessel. The same disposition of the
electrode is used as heretofore described.
Immediately the current is made, a series of bubbles of hydrogen
will run through the vessel which presently becomes pale and empty,
as a result of the electro-chemical action.
The needle should be allowed to remain in the vessel from five to
ten seconds, according to the size of the latter.
The object is to set up sufficient irritation in and of the walls of the
vessel so as to occlude it when cicatrization has been established.
Some edema follows such a treatment, subsiding in a day or more.
Several vessels may be treated in the same sitting, and at either side
of the face. The operator should guard against too strong a current,
to avoid scarring of the skin. The final result in this treatment shows
fine punctate scars, as after the removal of coarse hairs, and
sometimes pale linear scars, but these are observable only on close
inspection.

REMOVAL OF NÆVI
Birthmarks, port-wine marks, and other pigmentary conditions
may be entirely or partly removed from the skin of the face,
according to the size of the area treated and the nature of the case.
For this purpose the single needle attached to the negative pole is
hardly sufficient, unless the spot is exceedingly small, therefore a
bunch needle electrode is used. This electrode has a number of fine
steel needles set into it, as shown in Fig. 519.
In this treatment the needles are made to puncture the skin at
right angles to them to a depth corresponding to the papillary layer.
These pigments lie above that, so that it is not necessary to include
the derma. At each point of puncture a white spot will appear which
soon turns red. In a day’s time a number of fine scabs, or a single
scab, will form over the parts treated, which fall away in about five
 days eventually, leaving
the parts paler than
before, owing to a
number of minute
punctate scars.
The amount of
treatment given in each
Fig. 519.—Multiple Needle Electrode. case varies with the
extent of the lesion. If
the result from the first
sitting has not accomplished as much as desired, it can be repeated
over and over until the parts assume a normal tint. There may be
more or less bleeding following the treatment; this is easily checked
by pressure. If the part worked on is quite large, dry aristol dressing
should be used to avoid infection. The scab should not be picked off
by the patient, but allowed to fall off.

REMOVAL OF TATTOO MARKS

The best method of removing such pigmentations of the skin is to
remove them with the knife when possible, and to cover the wound
by sliding flaps made by subcutaneous dissection at either side of
the wound, as in the Celsus method. Some authorities advocate their
re-tattooing with papoid solution, while others prefer caustic agents,
with the object of destroying the pigmented area. These methods
are not to be preferred, since they leave unsightly burn scars.
Electrolytic needling may be tried and is quite successful when the
marks are very small, but, as with gun-powder stains, they are best
removed by punching, or cutting out, a little cone of skin containing
the pigment. The secondary wounds thus made leave only very
small punctate scars that are hardly noticeable. Of course a number
of such removals would not be advisable.
Where the pigmentation is very pale, recourse may be had to the
peeling method, as will be later described.
 THE TREATMENT OF SCARS
Not infrequently the cosmetic surgeon is called upon to remove or
improve unsightly scars about the face, the result of injuries or burns
and after the careless coaptation of such wounds. The scars vary in
extent and degree, from a mere pit due to varicella or variola to the
broad areas following the cicatrization of lupus and burns. Surgical
scars vary also from a mere line to areas of greater or less extent,
dependent upon the ablation of neoplasms or the granulation of
wounds due to any cause.
The treatment of scars depends upon their size and location. A
mere linear scar may be reduced by electrolysis, the needle,
negative pole, being introduced equidistantly, from one sixteenth to
a quarter inch apart, with the hope of causing a breaking down
electro-chemically of the scar itself and waiting for secondary
cicatrization. In other words, making a scar within a scar.
This mode of treatment may be repeated in two or three weeks
and has the tendency of breaking up the shiny line of light that
makes the scar stand out prominently from the skin.
Such scars, where nonadherent, or flat with the plane of the skin,
may also be tattooed to reduce their white color.
For this purpose, the red or carmine pigment used for tattooing is
diluted and pricked into the scar tissue with a fine cambric needle by
hand or electric process.
When the scar is small the line is punctured here and there and
the aqueous solution of the pigment is painted over the area, which
is again worked over to make it take.
For larger scar surfaces multiple needles are used. These are
composed of from four to ten needles soldered together at their eye
ends, leaving the points at an even level.
The electric method is the most serviceable for tattooing large
scars.
 These instruments are electro-magnetic devices made to
accommodate single or multiple needle points and can be obtained
from instrument makers.
The author has had a special electric synchronous reciprocal
apparatus made, as here shown in Fig. 520, which is much more
compact than the ordinary electric apparatus found on the market. It
works on the principle of the sewing machine needle.

Fig. 520.—Author’s Electric Apparatus for Tattooing Scars.

In using the electric apparatus the needle ends are dipped into the
pigment paste, to which a little glycerin is added to bind it, and this
is tattooed or pricked into the scar.
If, after the parts are healed, the color is too light, the scar may
again be gone over until the tint matches somewhat the tint of the
skin. Other pigments may be used, according to the complexion of
the patient.
Some scars, the resultant of negligent coaptation, are to be
excised according to the Celsus method and are brought together
with a number of fine silk sutures.
If the skin is found to be attached too closely to the subcutaneous
structure, it must be dissected up to render it mobile.
When the scar cannot be removed by excision the hypodermic use
of thiosinamin may be tried.
Thiosinamin or rhodallin is only slightly soluble in water, but the
addition of antipyrin according to Michel renders it useful for
hypodermic use. The formula preferred by the author is made as
follows:

℞ Thiosinamin grs. ij
Antipyrin grs. j
Aqua dest. gtts. xx.

The above solution makes up a single injection, which is to be

made directly under the scar or into the muscular tissue below it.
Two injections are given each week.
The treatment is to be continued until the texture of the cicatrix is
equal to that of the skin.
These injections are more or less painful and may be supplanted
to advantage with the hypodermic use of fibrolysin (Mendel), in
which each 2.3 c.c. correspond to three grains of thiosinamin.
For very small scars, as those occasioned by blepharoplastic
operation, the author employs the twenty-per-cent thiosinamin
plaster mull made by Unna. These are to be applied every day or
night, according to the convenience of the patient, and allowed to
remain on for several hours each day.
At first these plaster mulls are inclined to cause erythema and
exfoliation of the epithelium, therefore they might be used on
alternate days to keep the parts more sightly.
For scars of large extent the above method will answer best. If
there is considerable contraction, the parts should be massaged daily
to soften and stretch them. Eventually the depression of contour
may be corrected by hydrocarbon protheses introduced
subcutaneously following subcutaneous dissection, if deemed
necessary.
Small pits, where discrete, are best removed with a fine knife and
brought together by a fine suture which is to be removed on the fifth
day.
 Confluent pittings, as after variola, must be removed by
decortication or peeling methods.
The pits, if spread about the face promiscuously, may be treated
separately by the peeling method, but when they lie less than one
inch apart, it is best to treat the skin of the whole face.
This is done by applying pure liquid carbolic acid to the skin with a
cotton swab. The skin at once assumes a white color. If the pittings
are not very deep, one application of the acid is sufficient. If deep,
one or two more applications are made as the preceding one dries.
In very deep pits, the surgeon should apply the acid to the pit
proper several times, blending off the application at the periphery.
When the surface thus treated has become dry, adhesive plaster,
cut in half-inch strips of desirable length, are put on the face, one
above the other, slightly overlapping, until the whole treated surface
is well covered, mask-like.
The author uses Unna’s zinc oxide plaster mull for this purpose, as
it is backed with gutta-percha, which readily adapts itself to the
curvatures of contour.
The adhesive plaster mask is not removed until about the fourth
or fifth day, when it will be practically forced away from the skin by
the excretions thrown out from the derma. In some cases there is
considerable pus.
After removal of the mask the skin, now very red and tender, is
cleansed with a solution of bichloride, 1 in 10,000.
After the cleansing a mild soothing ointment, such as zinc oxide in
vaselin, is used for several days until the skin takes on its normal
epitheliar layer and appears normal in color.
No water or soaps are to be allowed during the latter period. In
the later days of the treatment the skin may be cleansed with a little
borated vaselin or even olive oil used with absorbent cotton.
If there is a pigmentation of the new skin this should cause no
alarm, as it will fade out in from six to eight weeks.
 Tincture of iodine has been used for the same purpose, as well as
its mixture with carbolic acid.
Resublimed resorcin is also advocated, but the resultant peeling
will not prove thick enough to give a satisfactory result.
If, for any reason, the effect obtained is not as desired, the patient
should wait for several weeks and have the treatment repeated.
It is hardly necessary to say that the application used should not
get into the eyes. The upper eyelids should not be treated, since no
benefit arises from it. If there is a redundancy of tissue, it should be
removed surgically, as heretofore described.
 CHAPTER XVIII
CASE RECORDING METHODS

Every case, whether of little consequence or of important nature,

should be properly and fully recorded in a thorough and systematic
manner. Apart from the value of such a record, to the operating
surgeon it often proves of the greatest importance in cases where
operations of a purely cosmetic nature are undertaken.
Patients who beg us to make them more beautiful, or less
unsightly in the eyes of the ever-critical observer, are the most
difficult to please, and often complain, after a few days of constant
mirror study, of the parts changed by methods that are the result of
years of hard-earned experience, that the nose or the eyes or the
ears have not been changed as much as they desired—in fact, so
little that their closest friends have failed to evoke ecstatic remarks
about the improvement.
This is not unusual with the most intelligent patients and is due to
the fact that cosmetic operations performed on an ugly though
otherwise normal organ have not yet become very frequent, and
while friends are inclined to remark a change in lesser defects, they
fail to credit this to the cause, owing to a lack of the knowledge of
cosmetic surgery, or their ignorance of the art entirely.
Photographs.—Where a pathological defect, wound, or scar or
traumatic deformity is to be corrected, the patient is usually kind
enough to permit of photographs being made of the parts to be
operated on, but where the defect is hereditary, or the result of age,
objections are invariably raised by all concerned, for fear their
pictures will be used in some outlandish way.
The objection to photographs is obvious, since it usually requires
visits to a studio, and the necessary loss of time to the surgeon,
whose presence is nearly always necessary to secure the proper
negative.
This is especially true of the nose. Very few photographers will
make a satisfactory sharp profile picture. It is less artistic, but most
desired by the surgeon, and when the patient is presented for a
second negative after the operation has been performed, the picture
varies more or less in pose from the first taken.
It would be well for physicians to have a camera for use in the
operating room, and those who can manipulate one will find that
taking a 5 × 7 negative the most suitable.
Stencil Record.—For those who cannot provide themselves or
bother with a photographic apparatus, the stencil record is
recommended.
For this purpose a picture of a normal eye and its lids, a nose, lip
or ear, is drawn upon a piece of oiled or stencil paper, or upon any
thick, stiff book board.
The paper is laid down upon a plate of glass and the outlines of
the picture are cut out, wide enough to allow the sharpened point of
a pencil to pass. Where the lines are long it is advisable to allow
connecting links to remain at various intervals as desired to keep the
stencil stiff and to prevent cut margins from slipping or rolling up.
(See Fig. 521.)
 Fig. 521.—Nose Stencil.

The stencil thus made is laid upon the record card and a tracing is
made upon the latter by passing the lead-pencil point along the cut
outline.
The stencil is now lifted and the defect sketched into the picture of
the normal organ.
If this should be the anterior nasal line, a perfect sketch can be
made of the defect by placing a card alongside of that organ and
drawing the outline upon it as the pencil is made to glide over the
nose, the point facing the card in such a way that a true profile
outline is obtained. The card is then cut along the pencil line.
The nasal section of the card is now placed upon the stenciled
nose and its outer border traced into or over it, as the case may be,
by drawing the pencil point along the outer margin.
The same method may be followed post-operatio. This method
can be employed for the other parts of the face as well, as, for
instance, the mouth, ears, base of nose, etc.
Distances in measurements should be put into the record drawing
to make it more exact.
The Rubber Stamp.—Another method is to make outline
sketches of normal parts of the face with India ink upon drawing
board and have those reproduced in rubber stamps, using the stamp
in place of the stencil and marking in the defect in the manner
before mentioned.
The Plaster Cast.—The best method by far, however, and the
one found most accurate, is the plaster cast. It is not a difficult thing
to make a cast of a nose, eyelid, lip, or ear, and the latter is much
more preferable to any other method of record.
For this purpose some modeling clay is required, which is molded
into a strip and laid around the part to be reproduced.
This forms a sort of raised ring or border and prevents the
overflow of the semiliquid plaster, and avoids the annoyance of
trickling the liquid upon other parts of the face about the site of the
part worked on; at the same time it permits of neatness and
uniformity in the size and shape of the casts to be filed away as
records. (See Fig. 522.)
Welcome to our website – the ideal destination for book lovers and
knowledge seekers. With a mission to inspire endlessly, we offer a
vast collection of books, ranging from classic literary works to
specialized publications, self-development books, and children's
literature. Each book is a new journey of discovery, expanding
knowledge and enriching the soul of the reade

Our website is not just a platform for buying books, but a bridge
connecting readers to the timeless values of culture and wisdom. With
an elegant, user-friendly interface and an intelligent search system,
we are committed to providing a quick and convenient shopping
experience. Additionally, our special promotions and home delivery
services ensure that you save time and fully enjoy the joy of reading.

Let us accompany you on the journey of exploring knowledge and

personal growth!

textbookfull.com