Advisor(s)
Abstract(s)
For multivariate non-stationary time series modeling is essential to know the number of common factors that define the behavior of the series. The traditional way to approach this problem is to study the cointegration relations among data through tests of the trace or maximum eigenvalue, obtaining the number of stationary long-run relations. Alternatively this problem can be analyzed using dynamic factor models as in Peña and Poncela (2006), estimating in this case the number of all independent common factors, stationary or not, that describe the behavior of data. In this context, we analyze empirically the power of such alternative approaches by applying them to series simulated using known factorial models. The results show that when there are stationary common factors, when the number of observations is reduced and/or when the series have involved more than one cointegration relation, the common factor test is more powerful than the usual cointegration tests. These results together with the greater flexibility of
dynamic factor models for identify the load matrix of the DGP make them more suitable for use in multivariate analysis.
Description
Preliminary version
Keywords
Multivariate analysis Common factors Cointegration Dynamic factor models Stationarity