Dollarization and foreign currency deposits in Latin America: A theoretical and empirical reassessment

To overcome the incompatibility between theoretical and empirical studies in the dollarization literature, a theoretical model, extended from Tobin's Portfolio balance model is derived to allow substitutability between domestic money and foreign currency deposits. In this way, the theoretical foundation for FCD demand is built and the justification for using FCDs in dollarization literature is achieved. Moreover, to enrich the explanatory ability of this model, risk and hysteresis variables are both built into it to reflect uncertainty of inflation and hysteresis effects on the money demand in a highly dollarized country, which has never previously been done in dollarization literature. Hysteresis in these asset demands is now directly observed and the determinants of M1 and FCDs are reassessed.; Four different estimation methods, (1) Nonlinear Least Square, (2) Cointegration Analysis (VEC), and (3) Generalized Method of Moments, (4) GMM System Estimation, are performed on the asset demand function of M1, real M1, FCDs and real FCDs using data on Argentina, Mexico, Peru and Uruguay. The results from these different estimations are very similar and consistent with the predictions of our theoretical model. In addition, based on consistent significance of risk and hysteresis variables in the regression results, these variables appear to be stronger determinants than conventional return variables. Therefore, in order to assess monetary asset demands correctly in dollarized countries, one needs to add these two types of variables to conventional portfolio theory to account for the different macro structure relating to inflation instability. This adds an important dimension to conventional portfolio theory. Moreover, three major contributions are made based on the following empirical findings: (1) money demands in dollarized countries show hysteresis effects of inflation and inflation variation, (2) similar hysteresis effects apply to FCD demands. (3) Risk and hysteresis variables are shown to be key determinants of asset demand and their inclusion in theoretical and empirical frameworks should be a necessity.