Due to limited access to foreign savings after the 2008 crisis, transition economies are forced to rely more on domestic savings in financing their growth. In that respect, it is often argued that the government should use tax policy to encourage domestic savings. Since the personal income tax reform is a burning issue in Serbia, the aim of this paper is to provide empirical evidence on the expected effects of each of the three income tax reform scenarios (flat, dual and comprehensive income tax scheme) on personal savings in Serbia, by taking into account both capital income tax effects and labour income tax effects. Taylor's theoretical model suggests that the personal saving is a function of personal income and the rate of return to savings. This is one of the seminal papers, in which the savings effects of tax policy reform are empirically estimated for a transition economy by taking into account both transmission channels. By combining Engle-Granger cointegration methods based on monthly macro data from 2004 to 2009, with the tax-benefit microsimulation model based on cross section micro data for 2007, it has been estimated that changes of capital income tax rate effects prevail over the effects of labour income tax changes, in terms of savings response. The results suggest that introduction of dual income tax in Serbia would boost personal savings in the long run, by 0.20%, while the flat tax and comprehensive income tax would lead to its decline by 2.15% and 3.64% respectively.