Money, transactions and the business cycle: inspecting the mechanism

This thesis analyses and compares two monetary models - the cash-in-advance (CIA) model and the real-resource-cost (RRC) model - using a Dynamic Stochastic General Equilibrium (DSGE) framework, with perfectly competitive markets and flexible prices. The CIA model is built on the quantitative assessment developed by Cooley and Hansen (1989, 1995), while the RRC model is an implementation of the analytical work by Feenstra (1986) and Wang and Yip (1992) on transaction costs. In order to inspect the mechanisms implicit in the monetary models at hand, this thesis analyses also some extensions, building on the seminal contributions by Stockman (1981) and Abel (1985). The main results emerging from the impulse-response functions are that the CIA and the RRC models respond in the same way to a shock in total factor productivity, while they differ in the propagation mechanism of the monetary shock, where the differences depend on the mix between cash- and credit-goods in the model economies. Instead, the impact of a transaction cost shock, in the case of the RRC model, remains weak. When compared with the stylised facts characterising the U.S. business cycle data, the CIA and RRC approaches exhibit the 'dichotomy' typical of the standard RBC literature: the volatility of real expenditure and working hours (and the respective correlation with output) are essentially driven by the technology shock, while nominal variables are mainly affected by the monetary shock. However, when it comes to the correlations of the endogenous variables with money growth, the CIA and the RRC models fail along many dimensions, when only consumption is linked with money (Chapter 3). By contrast, when transaction technologies are extended to investment and the money supply process is modified (Chapters 4 and 5), the empirical performance of the extended CIA model is superior with respect to the extended RRC model.