Document Type : Research Paper
Authors
1 Faculty Member of Statistics, Allameh Tabataba'i University, Tehran, Iran.
2 The Student of PhD, Statistics, Allameh Tabataba'i University, Tehran, Iran,
Abstract
The modeling is a very important topic in economic and financial research and it has a basic role in the analyzes, decisions, the policies and planning. In the modeling, assumptions have an important role in estimation and forecasting, because they can affect the results of models and analyses. The one of the most widely used classical time series models is the autoregressive model, where the current values are the finite linear combination of its past values. On the other hand, in real problems, many variables affect each other. For this reason, the vector time series models are used, which are part of the multivariate time series. The Vector autoregressive models are used in economic and financial modeling. The vector autoregressive (VAR) models are usually considered with the normal distribution for the shocks (noises). Since, in economic and financial issues, especially macroeconomics, the shocks don’t have symmetric distribution. In this paper, the VAR model with the Multivariate Skew Normal (MSN) distribution for the shocks is considered and since, the estimation of the parameters is an important step in modeling, the parameters of the model are estimated by using the Expectation Conditional Maximization (ECM) algorithm. Finally, by using the real data sets of Canada and Iran where the shocks have skewness and the evaluation criteria of the models, it is shown that the VAR model with MSN distribution for shocks in these data is more efficient than the VAR model with the multivariate normal distribution for shocks.
Keywords
- Vector autoregressive
- skewness
- multivariate skew normal
- maximum likelihood estimation
- expectation conditional maximization algorithm
Main Subjects