This paper aims to analyse the cointegration and causality relationships between inflation, GDP and unemployment by using Markov Switching - VAR and Markov Switching Causality tests for the period from 1957(2) to 2014(3) in USA. This study complements previous empirical papers. But at the same time, it differs from the existing literature by using Markov Switching VAR and Markov Switching Causality method which determined there is long-run relationship between inflation and unemployment for USA. Different MS-VAR models were estimated and the best model was selected based on AIC and LR test. When the transition probabilities are taken into account, important asymmetries in inflation, GDP and unemployment were recognized. The changes in the behavior of the variables were detected with MS-VAR models. The results highlighted the importance of economy policy on economic growth. (C) 2016 The Authors. Published by Elsevier B.V.