Abstract
Observing the economical changes and determining the factors related to financial failure are important for both the economical development of the country and for the self - evaluation of individual firms. In this study, the calculation of the financial success pobabilities for the Turkish firms in service sector and the investigation of temporal change in these probabilities are aimed. With this purpose in mind, financial stataments that are collected from İstanbul Stock Exchange are investigated, and multilevel statistical models are used for analysing this complex structured data which consists of repeated measurements in time. Specifically, Marginalized Transition Random Effects Models (MTREM) are fitted. By these models, it is possible to calculate financial success probabilities for each company, to compare success of companies in different subgroups, and to observe the changes in time. With a purpose of comparison, popular single level logistic regression models are fitted as well. In terms of the true classification rates, it is observed that MTREM is superior to logistic regression models.