Acta Universitatis Danubius. Œconomica, Vol 15, No 3 (2019)

Abstract

In modeling practice scholars and researchers use mathematical forms fitting to the data gathered from systems to judge the aforementioned assumptions. Econometric models in general can result in the explanatory opened expression in the sense that the set of variables involved in equations depend on the concrete economies. So, using currency demand approach (CDA) or MIMIC models in our calculation for informal economy in the Republic of Macedonia for a short period [2004, 2015] we faced the problem of correctly identification of variables that define the latent variable under study. To challenge this problem we used factorial analysis, partial least squares and monotonic correlation aside of recently referred analytic steps involving Empirical Mode Decompositions and log-periodical approach. This helped us to minimize the effect of small numbers of the data points, and non-stationary of variables in the procedure of fitting to the linear models. Therefore we evaluated the number of components in linear decomposition of the indicator of latent variable, most appropriate variables to be used in the forms and the number of possible joined latent variables to the informal economy