Deposit dollarization dynamics in a model of network externalities and hysteresis: the cases of Argentina and Bolivia

The shift from a domestic currency (peso) to dollar denominated deposits (deposit dollarization) in Argentina and Bolivia is explained with a nonlinear model that incorporates ratcheting effects, hysteresis and network externalities. There are two principal factors behind these effects: holding and transacting costs. The depreciation rate of a local currency versus an international monetary standard (peso versus dollar) is a reasonable proxy for the holding cost of the local currency. Strong depreciation of a currency imposes a substantial holding cost on its users and increases or "ratchets up" how high agents predict future depreciation rates will be. A transaction cost is incurred whenever an agent transacts outside his/her currency network of buyers and sellers. The model assumes there are two currency networks; one is based on the peso and the other on the dollar. A transaction network externality arises when relative returns or savings to using a given currency increase as the size of that currency's network of users expands. This externality makes the currency choice of an individual partially dependent on the choices others in his/her reference group or network make. Hysteresis, the persistence of high dollarization, happens when the expected holding cost of the dollar is less than the peso due to ratcheting effect, and/or converting back to the peso imposes a substantial transaction cost due to the large size of the dollar network. Under certain network dynamics and behavioral assumptions, dollarization levels can follow a nonlinear path to rest at an either high or low-equilibrium point. Resting at either high or low dollarization equilibria depends on the presence and location of an unstable third intermediate equilibrium point. A shock to the holding cost of at least one currency is needed to establish a new equilibrium position. For example, a spike in peso's depreciation ratchets up its expected holding cost and establishes a new nonlinear path that leads to higher dollarization equilibria, which the model expects the economy to start moving towards. The opposite outcome can happen if a strong contractionary policy appreciates the peso and reduces its holding cost.