Syllabus:
COURSE DESCRIPTION
This course provides an introduction to application of statistical modeling tools to problems in economics and business. It gives a hands-on experience with econometric modeling using SAS statistical software. The course covers statistical methods essential in economic modeling for analysis and policy. It covers single equation models, multiple equation models, models with discrete or categorical variables, parametric and non-parametric models, linear and non-linear models, models with time series data (stationary and non-stationary time series models, both single and multiple time series models), models with time lags, etc. The course seems to cover a lot of ground. In fact it does! But my emphasis is on econometric perspective, econometric analysis, and effective communication of the model and its results to the economists and public. However, theoretical results and derivations are also needed and the student is expected to cover those from the text book. Hands on experience through examples and computer software are provided during the class.
LEARNING OBJECTIVES
As a result of completing this course, you will get econometric perspective, knowledge of econometric methods, and effective communication of the usefulness of statistical applications to economics and business. In particular the course objectives are:
1. Understand the basic economic and business context in which economic data arise. In other words you will understand the Data Generating Process behind economic and business data.
2. Understand the importance of data warehousing and data mining
3. Understand the specific statistical modeling issues arising in economics.
4. Review a wide variety of statistical models so that a good model can be chosen from them.
5. Choose among alternate economic models using statistical criteria of goodness of fit.
6. Be able to construct credible statistical models to fit the given economic data
7. Understand and communicate to others how statistical modeling helps business, industry and government.
COURSE PREREQUISITES
The course presumes knowledge of probability theory, basic statistical inference, and statistical theory and methods of linear models (introductory).
TEXTBOOK (Required)
Developing Econometrics, First Edition (2012), by Hengqing Tong, T. Krishna Kumar, and Yangxin Huang, John Wiley and Sons Limited,
ISBN: 978-0-470-68177-0
Library Catalog # HB 139.T66-2011
330.01’5195-dc 23
This book can be obtained from GW Book Store or from eFollett.com.
This book is referred to in the syllabus as TKH.
Additional course material will be posted on Blackboard
The course uses the SAS. The students are required to use SAS for class use. The book uses DASC software (Data Analysis for Statistical Computing) developed by my co-author Professor Hengqin Tong. As many of the students in the class are Chinese they may find it useful to see the documentation and use of DASC. The web sites for the book and software maintained by Professor Tong are:
Chinese Language: http://tonghq.scholar.whut.edu.cn/
English Language: http://tonghq.scholar.whut.edu.cn/English/index.HTM
Download the DASC software from: http://tonghq.scholar.whut.edu.cn/English/Login.htm
GRADING
1. 20% for Home Work. The graded Home Work Assignments appear with an asterisk sign in the syllabus.
2. 10% for classroom presentation of the Home Work Assignment.
3. 20% for Midterm Examination (Closed Book) with SAS applications in the computer lab.
4. 25% for Term Paper (involves choosing a suitable problem in economics, collecting data, and applying one or more econometric techniques learned in the course). Term Paper is due on December 5 by 11PM electronically.
5. 25% for Final Examination. This Final Examination will consist of Part A: theoretical part (closed book) and Part B: computer applications in the computer lab.
The term paper topic should be chosen before Thanks Giving Holidays so that the Term Paper can be completed before December 10. No additional time will be allowed under any circumstances.
SCHEDULE
Holidays: Thanksgiving on November 26
Midterm: a two-hour midterm exam will be given on October 8 (3-5 PM) during the class hours. It is a closed book examination.
Final Quiz (Closed Book) is on December 10 in Room 602.
Term Paper is due on December 5 by 11 PM electronically.
SESSION-WISE LIST OF TOPICS
HOME WORK ASSIGNMENTS APPEARING WITH * ARE DUE FOR SUBMISSION FOR GRADING BEFORE THE NEXT CLASS
SESSION 1: AUGUST 27
Introduction to the Course and Syllabus
1.1 Nature and Scope of Econometrics
1.2 Types of Economic problems and types of economic data and types of econometric models
1.3 Data preparation: Cleaning and editing data, reprocessing of data, exploratory data analysis, pattern recognition and data mining.
1.4 Econometric modeling process
1.5 The need for graphics and effective presentation of results in business analytics
Assignment/Reference Material: TKH: Chapter 1.
Computer Applications: Multiple Regression, SAS Procedure REG
Home Work1*: Data Set1-Factors Affecting the Profits-Illustration of Exploratory Data Analysis and Model Choice
Home Work2: Data Set2-BSE Data of Paras Singhal: Illustration of the effect of Data Reprocessing before Analysis
SESSION 2: SEPTEMBER 3
2. Brief Review of Linear Multiple Regression Model
2.1 Brief review of standard linear model
2.2 Selection of independent variables and criteria for selection: Adjusted R2, Variance of prediction errors, Mallows’ CP statistic, Stepwise regression
2.3 Data transformations
2.4 Multicollinearity and Ridge Regression
2.5 Principal components and data reduction
Assignment/Reference Material: TKH: Chapter 2.
Computer Applications: SAS Procedure REG with options RIDGE, Ridge Trace, Variance Inflation factor (VIF).
Home Work3: Data Set3-Factors Affecting CEO Perks-Exploratory Data Analysis: Illustration of Selection of Independent Variables and Model Choice
Home Work4*: Data Set4-Multicollinear Data and Ridge-Regression: A Simulation Exercise.
SESSION 3: SEPTEMBER 10
3.General Linear Model
3.1 Heteroscedasticity: Consequences and tests for its presence
3.2 Weighted Least Squares
3.3 GLS
3.4 Autocorrelation as a special case
3.5 Fixed Effects, Random Effects, and error Component Models
Assignment/Reference Material: TKH: Chapter 3 (Omit Section 3.3.3, Omit 3.5.4 MINQUE)
Computer Applications: SAS Procedure REG with AUTOREG, BPG Test, SAS Procedure VARCOMP
Home Work5*: Data Set5:Heteroscedasticity Data: Testing for heteroscedasticity and correcting for heteroscedasticity.
Home Work6: Data Set6: Panel Data on Indian Banks: Error Component Model
SESSION 4: September 17
4. Regression Models with Discrete or Categorical Data
4.1 Discrete independent variables
4.2 Applications of dummy independent variables to regime shift, seasonal adjustment, etc.
4.3. Discrete Dependent Variable Models
4.4 Logit, Probit, Tobit, and General Burr Distribution Models
4.5 LR based tests for inference
4.6 Application of discrete dependent variables models
Assignment/Reference Material: TKH: Chapter 4 (Sections 4.1 and 4.2)
Computer Application: SAS Procedures LOGISTIC, PROBIT, QLIM, and LIFEREG.
Home Work 7*: Data Set 7 on Home Ownership from BWSSB Survey and Data Set 7B on Credit Card Default.
SESSION 5 September 24
5. Nonlinear Regression Models
5.1. Nonlinear Regression Models
5.2 Types of non-linearity and Growth Curve Models and Survival Models
5.3 Methods of estimation and Marquartdt’s algorithm
5.4 Tests of Hypotheses in Nonlinear Regression based on LR Test
5.5 Bootstraps and inference in Small Samples
5.6 Application of Nonlinear Regression: CES Production Function
Assignment/Reference Material: TKH: Chapter 4, Section 4.3 and 4.4 (Omit Section 4.4.4 and beyond).
Computer Applications: Procedures NLIN, QLIM and NLMIXED.
Home Work 8: Any one of the following two examples. Data Set 8A : Mexican Frontier Production Function. Data Set 8B: Engel Curve Estimation for Cereal Consumption.
SESSION 6: October 1
6. Nonparametric and Semi-parametric Models
6.1 Nonparametric density function
6.2 Nonparametric regression
6.3 Varieties of non-parametric regressions and the nearest neighbor regression
6.4 Stochastic Frontier Regression Models
6.5 Inference in Stochastic Frontier Regressions
6.6 Application of Frontier Regression: Stochastic Frontier Production Function
Assignment/Reference Material: TKH: Chapter 5 (Omit Section 5.3.1)
Computer Application: Procedure KDE for Kernel density estimation, Procedure TPSPLINE and Procedure LOESS.
Home Work 9A: Data Set 9A: Application of KDE Procedure with simulated data.
Home Work 9B: Data Set 9B: Cereal Consumption Data for India: Procedure TPSPLINE for nonparametric estimation of Engel Curve.
Home Work 9C: Using Data Set 9B: LOESS Procedure for estimating a non-parametric regression of Engel Curve for cereal consumption.
SESSION 7: October 8
Review of Topics Covered until then (2-3 PM)
MID-TERM EXAMINATION: CLOSED BOOK EXAMINATION (3:30-5 PM).
SESSION 8: October 15
8. Simultaneous Equation Systems:
8.1 Stochastic regressors in linear models and consequences
8.2 Simultaneous Equations Model and Economy-wide Models
8.3 Identification of a model
8.4 Simultaneous equation bias and bias of OLS estimator in a simultaneous case
8.5 Consistency of estimators
Assignment/Reference Material: TKH: Chapter 6 (Sections 6.1)
Home Work 10*: Data Set 10 Food market data set. Food Market Demand and Supply Models using Procedure SYSLIN Procedure with OLS and 2SLS options.
SESSION 9: October 22.
9. Methods of Estimation: Simultaneous Equations
9.1 Method of Instrumental Variables
9.2 Single equation and System-wide estimation methods
9.4 Small and large sample inference
9.5 Bootstrap methods for small sample inference
9.6 Application to a simple Macro econometric Model
Assignment/Reference Material: TKH: Chapter 6 (Sections 6.2)
Computer Application: Procedures SYSLIN and SIMLIN
Home Work 11: Data Set 11: Klein Data Set. Estimation and Simulation of Klein Model using SYSLIN and SIMLIN procedures of SAS.
SESSION 10: October 29
10. One Variable Stationary Time Series Models
10.1 AR and MA models
10.2 ARMA Models
10.3 Identification of an ARMA model
10.3 Estimation of ARMA Models and Box-Jenkins Approach
10.4 Forecasting using ARMA models
10.5 Inference in Stationary Time Series Models
10.6 Auto Regressive Conditional Heteroscedastic (ARCH) Model and Generalized Autoregressive Conditional Heteroscedasticity (GARCH) Model
10.7 Fitting an ARIMA model to the US GDP Series
Assignment/Reference Material: TKH: Chapter 7 (Omit Section 7.2.3 on Spectral Density. Go back to Section 3.3.3 of Chapter 3)
Computer Application: SAS ETS Procedure ARIMA
Home Work 12: Data Set 12: Simulated AR, MA, and ARMA Processes data. Fitting AR, MA, and ARMA Models.
SESSION 11: November 5
11.0 Non-stationary Time Series Models
11.1 Dickey Fuller Test for Non-stationarity
11.2 Bootstrap test for Non-stationarity
11.3 Augmented DF Test
11.4 ARIMA Models and Box-Jenkins Approach
Assignment/Reference Material: TKH: Chapter 8, Sections 8.1 (except Sections 8.1.3-8.1.6), and Section 8.2.
Computer Application: ARIMA Procedure.
Home Work 13*: Data Set 13: Logarithm of US Real GDP. ARIMA Modeling.
SESSION 12: November 12
12.1 Finite and Infinite Distributed Lag Models
12.2 Auto Regressive Distributed Lag (ARDL) Models
12.3 Vector Time Series
12.4 Vector Autoregression (VAR) Model
12.5 Vector Moving Average Model
12.6 Granger Causality Test
12.7 Multivariate Non-stationary Time Series Models
Assignment/Reference Material: TKH: Chapter 6 Sections 6.3-6.5, Chapter 8, Sections 8.1 and 8.2
Computer Application: VARMAX Procedure.
Home Work 14: Data Set 14: Trade Deficit and Budget Deficit in USA. Modeling Causality
Session 13: November 19
13.0 Co-integration and Error Correction
13.1 Co-integration
13.2 Error Correction
13.3 Co-integration and Error Correction Models
13.4 Testing for the number of co-integrating equations
Assignment/ Reference Material: TKH: Chapter 8, Section 8.3
Computer Application: VARMAX Procedure
Home Work 15: Data Set 15: Stock Prices of Some of the BRIC Countries and their Co-integration with stock prices of developed countries.
SESSION 14: December 3
14.0 Dynamic Structural Equation System and Vector Auto Regression
14.1 Autoregressive Distributed Lag (ADL) Model
14.2 Structural Equation System Model
14.3 Vector Auto Regression Model
14.4 Synthetic representation of Structural Equation System
14.5 Impulse-Response Analysis
Assignment/Reference Material: TKH: Chapter 8, Sections 8.3.2, And Section 8.1.4
Computer Application; VARMAX Procedure to compute Impulse-Response relationship.
Home Work 16: Data Set 15: Stock Prices in some of the BRIC Countries and US and UK. Application of VAR-EC Model, using VARMAX to estimate relation between Stock Prices in different countries through Impulse-Response Functions.
December 10: Final Examination. This is a closed book examination.