Vector autoregressive model with multiple sets of exogenous variables
Loading...
Date
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
University of Cape Coast
DOI
Abstract
The VAR-X model designed for multivariate time series data which consists
The VAR-X model designed for multivariate time series data which consists
of a set of endogenous variables and a single set of exogenous variables will
require substantial extensions to meet data problems that involve multiple sets of exogenous variables. The study precisely provides such extensions to obtain the VAR-MX model. For ease of simplification, the new model is formulated for a general lag 𝑝 in both types of variables. In terms of well-defined notations, the study specifies the design matrix and derives the corresponding vector moving average (VMA-MX) for the model with an introduction of a matrix polynomial in terms of the lag operator. The maximum likelihood estimation of parameters is then recast for the problem. The new technique is applied in crop modelling that examines the amount of yield of four rice varieties using five climatic conditions and three soil characteristics as two sets of predictors. With an initial Box-Cox transformation of the data carried out at an optimal parameter value, an efficient model is obtained with negligible error measures which also addresses exacerbated issues of multicollinearity and serial correlation.
Description
xvi, 254p. :ill.
