–
- Code:
- ECOM-412/SOC-9412
- Field:
- Econometrics
- Targets:
- Master’s students Research Master's students PhD students
- Organiser:
- University of Helsinki - Economics
- Instructor:
- Jani Luoto
- Period:
- Period 1
- Format:
- Participation in teaching
- Method:
- Contact teaching
- Venue:
- Economicum
- Enrollment:
Equivalent to ECOM-412 Applied Macroeconometrics 2
In case of conflicting information consider the Sisu/Course/Moodle pages the primary source of information.
Aalto, Hanken and UH economics students can enroll through their home university’s SISU. Further instructions are available on the How to enroll? page, also for students from other universities.
If you would like to count the credits towards your degree, please check your curriculum or contact your supervisor or student services for guidance.
- To access the Moodle course area, use all the features and participate in the activities (assignments, discussions), you must have successfully registered for the course in Sisu and logged in with your UH user ID.
- For more information on how to activate your UH user ID and register for a Moodle course area, click here.
Content
This course introduces Bayesian analysis of vector autoregressive (VAR) models. The first part of the course focuses on reduced-form VARs, covering both the standard homoskedastic Gaussian VAR and VARs with stochastic volatility, which are widely used as workhorse forecasting models in central banks and other policy institutions. Different prior specifications are discussed, along with predictive inference.
The second part of the course turns to structural VAR (SVAR) models, where the goal is to draw causal inferences about the effects of economic shocks. Topics include set identification, sign restrictions, and statistical identification based on stochastic volatility and non-Gaussianity. The course emphasizes how different identification strategies shape causal interpretation, combining theoretical foundations with practical implementation to equip students with the tools to evaluate and apply alternative approaches to causal analysis in time series.
In addition, the course introduces computational techniques for Bayesian inference, ranging from standard numerical integration methods such as Gibbs sampling to Hamiltonian Monte Carlo (HMC). HMC has become a state-of-the-art tool for efficiently sampling from high-dimensional posterior distributions and is particularly well suited for modern Bayesian time series applications.
The emphasis throughout the course is on the application of these methods to macroeconomic time series data, providing students with the tools to carry out their own Bayesian VAR and SVAR analyses in practice.
Learning outcomes
After the course, the student should
- Be familiar with Bayesian methods for reduced-form vector autoregressive (VAR) models, including standard prior specifications, models with stochastic volatility, and predictive inference.
- Understand how VARs are used as forecasting tools in applied macroeconomic research and policy institutions.
- Know the main approaches to identifying structural VAR (SVAR) models including sign restrictions, and statistical identification based on stochastic volatility and non-Gaussianity.
- To understand the concept of set identification and its implications for structural VAR analysis.
- Be able to distinguish between reduced-form and structural analysis, and understand their role in causal inference with time series data.
- Be familiar with computational techniques for Bayesian inference, including Gibbs sampling and Hamiltonian Monte Carlo (HMC).
- Be able to apply these methods to macroeconomic time series data, and to conduct their own Bayesian VAR and SVAR analyses.
- Be able to present and interpret empirical results obtained with the methods covered in the course.