Divisia Monetary Aggregates and Exchange Rate Forecasting

Abstract

Divisia monetary aggregates have been shown to be an improvement on the simple-sum monetary aggregates used by policy makers in the great majority of central banks in the world. Since Barnett (1978, 1980) derived the User Cost Price and produced the theoretically correct from of aggregation, they have helped solve some of the difficult problems in the profession. One such problem is forecasting exchange rates. SinceMeese & Rogoff (1983) convincingly argued that no model could outperform a driftless random walk in predicting exchange rates, there have been many papers which have tried to find some forecasting methodology that could beat the random walk, at least for certain forecasting periods. In particular, Wright (2008) introduced Bayesian Model Averaging as a tool to forecast exchange rates and Lam et al. (2008) compared Bayesian Model Averaging and three structural models to a benchmark model (the random walk), both studies obtaining positive results. Also, Carriero et al (2009) found positive results using a Bayesian Vector Auto-regression model. Barnett & Kwag (2005) availed themselves of the User Cost Price and Divisia monetary aggregatesandincludedthemasvariablesintheFlexiblePriceMonetarymodel, Sticky Price and Hooper Morton models to show that it has greater forecasting power than the random walk when the aforementioned variables replace the interest rate and simple sum monetary aggregates (respectively). Specifically, they worked with the US dollar/British pound exchange rate. This dissertation aims to extend three different experiments. The first chapter compares Purchasing Power Parity, Uncovered Interest Rate, Sticky Price, Bayesian Model Averaging, and Bayesian Vector Auto-regression models to the random walk benchmark in forecasting exchange rates between the Paraguayan Guarani and the US Dollar, the Brazilian Real and the Argentinian Peso. The second, follows Barnett and Kwag’s work, applied to the US dollar/Euro exchange rate, but also includes an Uncovered Interest-rate Parity model. I use the Root Mean Square Error, Direction of Change statistic, and the Diebold-Mariano statistic to compare the forecasting power of the models in Chapters 1 and 2. In the first chapter, resultsindicatethatinshorterforecastinghorizonsBayesianModelAveraging,and Bayesian Vector Auto-regression models perform better than the random walk and that structural models outperform the random walk at longer horizons. In the second chapter, results indicate that Uncovered Interest-rate Parity with the User Cost Price instead of the interest rate improves on the random walk forecasts in every time horizon. In view of the results in Chapter2, Divisia monetary aggregates for the country of Paraguay are calculated in the last chapter with the aim to examine their performance against simple-sum monetary aggregates in estimating money demand. Results suggest that Divisia monetary aggregates are superior to simple-sum aggregates in money demand estimation.