turkmath.org

In today’s world, it is vital to foresee the changes in relations between macroeconomic variables to be pre-active by developing appropriate mitigation policies in the case of emerging global crisis. Therefore, constructing early warning systems becomes crucial in a global scale. This requires observing cross sectional data over time, named as longitudinal or panel data. Although the aim of panel data analysis is to model differences in behavior across individuals, ours is to capture shifts over time in macroeconomic relations constructed by the use of cross sectional data. However, adjusting these techniques to construct early warning systems is not so direct since structural break tests can detect the changes after they occur which is too late. Thus, we establish the relation between macroeconomic variables by linear regression method at each point in time separately. Then, to observe the behavior over time for early warning global crisis, we introduce the linear profile control charts, a tool newly being developed in statistical process control. In these charts, it is assumed that the explanatory variables’ values do not change from profile to profile and there is no correlation between profiles. However since we are dealing with economical and finance data both of these assumptions become invalid. Thus we adopt these charts to take into account both violations and see whether the global crisis can be detected before they occur. For this purpose two pilot studies are performed. One models the relation between GDP and energy consumption while the other models the relation between GDP and stock exchange and tries to capture the crisis by monitoring these relations with the use of two different control charts. The appropriateness of them is discussed as well as the comparison of their performances.