The objective of this thesis is to examine the effect of pension system design on saving. Chapter II analyses the relationship analytically, in a two-period two-generation partial equilibrium Samuelson-type of OLG model in the logarithmic format, augmented by design of the public pension system and by the mandatory funded pension programme with the displacement coefficient. The model predicts higher household saving for countries with a lower contribution rate, higher redistribution within the public system and greater importance of private pension savings, i.e. systems that could be classified as ‘Beveridge’. Partial derivatives of the model are numerically simulated. Chapter III first deals with the measurement issue, defining the set of ‘pension design indicators’ that will be used later for the empirical analysis. Then it tests the ‘convergence hypothesis’ of pension models using several methods. The results unambiguously suggest that, despite a convergence in pension policy goals, convergence of pension models has not occurred i.e. the pension systems around the world are still influenced by their historical paths. Chapter IV empirically investigates the effect of pension system design on saving rates. The first part of the analysis closely follows work in Disney (2005), with somewhat differently calculated public pension design variables and with data for the 2000s. I also tested the impact of private pension component on household saving. The overall pension system design was estimated using principal components composite indices. The results obtained using a number of estimation methods have not confirmed the predictions of the theoretical model, and are actually counter-intuitive. In addition to methodological issues related to household saving data, a possible explanation for this could be the complexity of household saving behaviour that needs to be adequately controlled for. The final chapter summarizes the findings, discusses limitations of empirical investigation and sets forth directions for future research.