Business Knowledge Series course

Presented by Dr. Oral Capps, Jr., executive professor and co-director of the Agribusiness Food and Consumer Economics Research Center and holder of the Southwest Dairy Marketing Endowed Chair in the Department of Agricultural Economics at Texas A&M University

This sequel to Introduction to Applied Econometrics focuses on intermediate and advanced topics in working with econometric models. This course enables analysts to better understand their economic/business landscape and to improve their ability to make sound forecasts. Through applications, participants gain knowledge of the practical elements of applied econometric analysis. The overall aims are to sharpen the quantitative, statistical, and analytical skills of participants in dealing with problems and issues related to business and economics as well as to improve communication skills in reporting findings to decision makers.

Learn how to

• Detect and circumvent collinearity and ill-conditioning problems in econometric models.
• Detect and assess data outliers and leverage points.
• Detect structural change and test the stability of structural coefficients.
• Incorporate dynamic elements in econometric models principally through the use of distributed lags.
• Use Autoregressive Conditional Heteroscedasticity (ARCH) and Generalized ARCH (GARCH) models.
• Use qualitative choice models and censored response models.
• Use simultaneous-equation models.
• Use seemingly unrelated regression models.
• Use panel data in econometric applications (self-study).

Who should attend

Academicians, economists, forecasters, and government and business analysts

Formats available	Duration
Classroom:	3.0 days

Before attending this course, you should:

• Have a basic knowledge of SAS software, including SAS procedures such as PROC REG, PROC AUTOREG, and PROC MODEL.
• Know the equivalent of the material covered in the Introduction to Applied Econometrics course, specifically data issues inherent with econometric models, the development and estimation of single-equation econometric models, hypothesis testing associated with these models, the construction and interpretation of dummy variables, and the detection and circumvention of serial correlation (autocorrelation) and heteroscedasticity.
• Have some knowledge pertaining to developing and evaluating ex-post and ex-ante forecasts.

This course addresses SAS/ETS software.

Detecting and Circumventing Collinearity or Ill-Conditioning Problems

• Introduction.
• Collinearity diagnostics.
• Solutions to the collinearity problem.
• Examples.
• Commentary.

Detecting and Assessing Data Outliers and Leverage Points

• Background.
• Influence diagnostics.
• Solutions to the problem of influential observations.
• Robust regression techniques.
• Examples.
• Commentary.

Detecting Structural Change and Testing for the Stability of Structural Coefficients

• Introduction.
• Diagnostic tests for structural change.
• Example: US gasoline consumption 1960-1995.
• Illustration of sequential Chow tests.
• Illustration of the Farley, Hinrich, and McGuire test.
• Illustration of recursive coefficients, recursive residuals, CUSUM, and CUSUMSQ tests.
• Commentary.

Incorporating Dynamics through the Use of Distributed Lags

• Introduction.
• Approaches to distributed lag models.
• Sample problem: free-form lag.
• Sample problem: geometric lag.
• Sample problem: polynomial distributed lag.
• Partial adjustment model.
• Sample problem: partial adjustment model.
• Commentary.

Autoregressive Conditional Heteroscedasticy (ARCH) and Generalized ARCH (GARCH) Models

• Introduction.
• ARCH(q) model.
• Sample problem: ARCH(q) model.
• GARCH model.
• GARCH-M model variations (linear, square root, and log).
• GARCH model with autoregressive errors.
• Example: ARCH/GARCH models for three month returns of SP500.
• Threshold GARCH model.
• Exponential GARCH (EGARCH) model.
• Example: exponential GARCH (EGARCH) model.
• Estimation of GARCH models.
• Commentary.

Qualitative Choice and Censored Response Models

• Limited dependent variables.
• Probit/logit models.
• Computational methods and statistical considerations for empirical analysis.
• Sample problem: use of probit analysis.
• Sample problem: use of logit analysis.
• Censored response models.
• Censored samples: use of the Tobit model.
• Sample problem with the Tobit model.
• Heckman sample selection procedure.
• Sample problem with the Heckman sample selection procedure.
• Commentary.

Simultaneous Equation Models

• Modeling approaches.
• Simultaneous systems.
• Simultaneous structural models.
• Types of structural models.
• Identification issues.
• Example of order and rank conditions.
• Common methods of estimation.
• Simultaneous equation model of demand and supply relationships.
• Analytically derived reduced forms.
• Microeconomics specification of simultaneous equation models.
• Final form of the system.
• Determining whether a system is stable.
• Example of stability condition: Klein model.
• Specification, estimation, and simulation of a dynamic macroeconomic simultaneous equation model.
• Commentary.

Seemingly Unrelated Regression Models

• Seemingly unrelated regression models.
• Example of seemingly unrelated regression models.
• Example: demand for a cereal product from five retailers: HEB, Publix, Food Lion, Fred Myer, and Meijer.
• Seemingly unrelated regression models with restrictions.
• Rotterdam model.
• Linear approximate almost ideal demand system (LA/AIDS) model.
• Example: demand interrelationships for spaghetti sauces: LA/AIDS model.
• Demand for spaghetti sauce example: use of PROC MODEL.
• Commentary.

Pooling of Time-Series and Cross-Sectional Data

• Introduction.
• To pool or not to pool?
• Single-equation model specification associated with the pooling of time-series and cross-sectional data or the use of panel data.
• Typical assumptions when dealing with the pooling of time-series and cross-sectional data: Parks model.
• Typical assumptions when dealing with the pooling of time-series and cross-sectional data: error components model.
• Sample problem: the use of pooled OLS, Parks procedure, and error components procedure.
• Typical assumptions when dealing with the pooling of time-series and cross-sectional data: covariance model.
• Fixed and random effects.
• Sample problem: the use of the ANACOVA or LSDV model and the use of the one-way random effects model.
• Seemingly unrelated regression model (fixed and random effects).
• Sample problem: the use of seemingly unrelated regression.
• Commentary.