Course content

Topic overview

Part 1:

• Classical linear multiple regression models. Ordinary Least Squares (OLS). Hypothesis testing and evaluation of models. The classical OLS-assumptions, and deviations from these. OLS-regression with time series. Autocorrelation. Stationary and nonstationary variables. Spurious regressions.
• Panel data models. Fixed effects and random effects models. Hausman test. Tests for autocorrelation and heteroskedasticity in panel data.

Part 2:

• Time series analysis. Basic assumptions. Stationarity and autocorrelation.
• Estimation and interpretation of autoregression models - AR models.
• Tests for stationarity: Dickey-Fuller Tests.
• Residual analysis. Autocorrelation tests: Ljung-Box test. Tests of heteroskedasticity: ARCH test. Tests of normality.
• Estimation and interpretation of VAR (Vector autoregression models) and VECM (Vector error correction models).
• Cointegration. Engle-Granger test of cointegration.
• Interpretation of Granger causality in VAR and VECM models.

Part 3:

• Principal components analysis. Eigenstructure of the covariance matrix.
• Exploratory factor analysis. Factor analysis techniques. Factor rotations. Factor analysis versus principal components analysis.
• Simultaneous Equations. The bias of OLS regression. Two-Stage Least Squares 2SLS.

Part 4:

• SEM: Structural Equation Modeling with observed variables.
• Confirmatory factor analysis CFA. Path diagrams. Estimation and model evaluation. Applied examples with modern statistical software.
• Structural equation models with latent variables. Estimation with: ML, GLS and ULS.
• Model evaluation. Goodness of fit measures.
• Statistical assumptions for Structural Equation Modeling

• Part 5:
• Validity and reliability in SEM.
• Missing data in SEM.
• Nonnormality in SEM. Kurtosis and skewness. Robust estimation with RML and WLS.
• SEM with ordinal variables. Estimation with DWLS.

Applications of multivariate Statistical Methods and Structural Equation Models in economic analysis, using modern statistical software.

The statistical software that will be used during the course will be announced at the start of the semester.

Reserve the right to make changes in the academic content.

Learning outcome

Knowledge

The student has:

• In-depth knowledge in multivariate Statistical Methods and Structural Equation Modeling SEM, with good insight into estimation and evaluation of different types of SEM models, with both observed and latent variables. Students can apply different estimation techniques for non-normality and at different measurement levels of the variables.
• In-depth knowledge of estimation of multiple regression models and consequences of deviations from the classical assumptions for OLS. Students also work with time series models: AR-, VAR- and VECM-models.

Skills

The student can:

• Estimate and interpret various factor models, regression models, and Structural Equation Models using modern statistical software. Present the analytical results in writing.
• Understand the assumptions and limitations on which these techniques are based.

General competence

The student can:

• Suggest and understand the results of modern multivariate statistical techniques that are increasingly used in business and research.
• Handle the general structure model and understand the importance of including measurement errors in econometric models.
• Use relevant statistical software for analysis of economic data and be able to expand the toolbox as needed.

Learning methods and activities

Lectures and group work, where students will use multivariate statistical methods and Structural Equation Modeling in economic analysis, using modern statistical software. A high level of activity is expected in both compulsory and voluntary exercises.

Compulsory assignments

• Obligatoriske innleveringer

Specific conditions

Recommended previous knowledge

Quantitative methods and econometrics

Required previous knowledge

None

Course materials

Literature will be announced at the start of the semester.

Credit reductions