Humboldt-Universität zu Berlin >> Wirtschaftswissenschaftliche Fakultät

 

Subproject C15

Structural Vector Autoregressive Analysis
 
Contact:
   
Head of Project : Prof. Dr. Helmut Lütkepohl
 
Tel:

+49 30 8978 9236

Fax:

Email:

hluetkepohl@diw.de
 

Address:

DIW Berlin and Freie Universität Berlin (from January 2012)
DIW Berlin
Mohrenstraße 58 10117
Berlin, Germany

 

Employees

 

 

Description


The risk associated with economic activities depends partly on the phase of the business cycle. For example, the risk of becoming unemployed is larger in a recession than in an expansion. Similarly the insolvency risk of firms depends on the phase of the business cycle. Therefore business cycle analysis is of central importance for economic risk analysis. Structural vector autoregressive (SVAR) models have been used extensively in that context. In particular, they have been used for investigating macroeconomic issues such as monetary and fiscal policy implications, labor market and capital market performance, etc. In SVAR models a major problem is to find convincingly identified shocks which are informative about the actual reactions of a set of variables to unexpected exogenous innovations. Unfortunately, economic theories and models often provide insufficient information to fully identify the shocks of interest. A number of alternative proposals have been made for solving this problem. They typically use economic theory and institutional restrictions for identifying the shocks. Classical identification restrictions impose constraints on the instantaneous or long-run effects of the shocks. More recent proposals use softer restrictions on the signs of the effects of shocks derived from economic theory or economic insights.

In the present research project I intend to explore the possibilities for assigning a larger role to the statistical data properties in the identification of shocks. In particular, exploiting changes in the volatility of the system may help in identifying shocks. Such an approach seems promising given the recent turbulent macroeconomic conditions in most parts of the world. It will be explored how special features of data generation processes that account, for instance, for changes in the volatility of economic variables can be utilized for identification. An important proposal in this context uses a Markov regime switching mechanism to model the changes in volatility. It allows the data to decide on the changes in volatility and has a number of advantages over other possibilities for modeling changes in volatility. It will be a main ingredient in the identification procedure for shocks in economic systems explored in this subproject.

 
Home | Projekte | RDC

Contact CRC649 | Impressum