Applications of Time Series in Finance and Macroeconomics



Journal Title

Journal ISSN

Volume Title



This dissertation contains three applications of time series in finance and macroeconomics. The first essay compares the cumulative returns for stocks and bonds at investment horizons from one to ten years by using a test for spatial dominance. Spatial dominance is a variation of stochastic dominance for nonstationary variables. The results suggest that for investment horizons of one year, bonds spatially dominate stocks. In contrast, for investment horizons longer than five years, stocks spatially dominate bonds. This result is consistent with the advice given by practitioners to long term investors of allocating a higher proportion of stocks in their portfolio decisions. The second essay presents a method that allows testing of whether or not an asset stochastically dominates the other when the time horizon is uncertain. In this setup, the expected utility depends on the distribution of the value of the asset as well as the distribution of the time horizon, which together form the weighted spatial distribution. The testing procedure is based on the Kolmogorov Smirnov distance between the empirical weighted spatial distributions. An empirical application is presented assuming that the event of exit time follows an independent Poisson process with constant intensity. The last essay applies a dynamic factor model to generate out-of-sample forecasts for the inflation rate in Mexico. Factor models are useful to summarize the information contained in large datasets. We evaluate the role of using a wide range of macroeconomic variables to forecast inflation, with particular interest on the importance of using the consumer price index disaggregated data. The data set contains 54 macroeconomic series and 243 consumer price subcomponents from 1988 to 2008. The results indicate that factor models outperform the benchmark autoregressive model at horizons of one, two, four and six quarters. It is also found that using disaggregated price data improves forecasting performance.