A new look for country/region size selection in the Global VAR approach: Theory and Monte Carlo simulation

The Global Vector Autoregressive modeling approach is often used by many public and private institutions, especially central banks, both in policy making and in academic research. The application of the approach in the field of macroeconomics offers many model alternatives depending on how the country/region dimension is formatted. Creating models and achieving results involves many different steps. Most intermediate operations and decision-making processes related to these operations are carried out decisively by the researcher. In proportion to the increase in the size of the country/region, there is both a serious transaction cost and the probability that the researcher will make individual mistakes in making these transactions increases. Planning of models is done on some economic justification. But it has not yet been adopted to use any statistical analysis. The purposes of the approach's use in the field of macroeconomics have evolved to explain the variables that are considered important in a country. Thus, the main aim of the study is to develop a statistical point of view that can be used to compare all possible model options that can be established through the data set studied within the scope of the research topic. One of the most common tools used to select the appropriate model from among statistical and econometric models is the Akaike information criterion. When the conceptual and theoretical foundations of the criterion are reviewed, features such as the use of the probability estimation method at most are compatible with the Global Vector Autoregressive modeling approach. Requirements, such as the size of the sample data related to the implementation of the approach, are also within the desired characteristics of the Akaike criterion. Despite its significant advantages, the Akaike information criterion, which has a number of disadvantages, had to be made resistant to the conditions in which it is sensitive. Thus, some adaptations are made to the Akaike-type information criterion to conform to the econometric theory of Global vector autoregressive models. The limited sample characteristics of both criteria derived were studied by Monte Carlo simulation experiments. In the design of the experiment, it is aimed to reflect various situations that can be encountered in the macroeconomic framework. As a result, a criterion has been obtained that can be used to determine the structure of the cross-section unit size in Global Vector Autoregressive models.