¿Es estable la demanda de dinero? Análisis del caso Uruguayo

We reestimate the demand for money during 1983.q1 – 2017.q1 for the uruguayan economy taking into account the current estimation carried out by Brum et. al. (2011). However, we introduce changes in how the relative volatility of returns for substitutes for money is modelled. We test for long term stability in the demand for money by using traditional cointegration tests. Furthermore, we analyse whether changes in the monetary policy and the exchange policy, the financial crisis experienced in 2002 as well as other factors not included in the cointegration relation have an impact on the stability of the demand for money. The long term stability in the demand for money is approached in three different, but complementary, ways. Firstly, we emphasize the use of Markov-switching models for which we define two different states for the demand for money, the probabilities of transition, and their expected duration. This approach is complemented with an analysis of recursive regressions and the study of connectivity between the relevant variables carried out through a vector autorregresive model (VAR) and the analysis of the decomposition of forecast-error variance decompositions using generalized variance descomposition. Unlike the break analysis or structural change models, the approaches used do not assume ex-ante the existence of change or instability. Furthermore, the approaches used do not assign, a priori, a date to the structural change or break in the series; the estimation carried out by using these approaches provide the date of such breaks or changes. This paper aims to show through the different methodologies, that although it is possible to find long term relationships through traditional cointegration tests, the data suggest that this relationship is unstable during the analysed period.