Event Details

This Seminar will be hosted by the Economics

Lorenzo Trapani will present “Determining the dimension of factor structures in non-stationary large datasets” (with M. Barigozzi)

ABSTRACT:

Wepropose a procedure to determine the dimension of the common factor space in alarge, possibly non-stationary, dataset. Our procedure is designed to determinewhether there are (and how many) common factors (i) with linear

trends, (ii) with stochastic trends, (iii) with no trends,i.e. stationary. Our analysis is based on the fact that the largest eigenvaluesof a suitably scaled covariance matrix of the data (corresponding to the commonfactor part) diverge,

as the dimension N of the dataset diverges, whilst theothers stay bounded. Therefore, we propose a class of randomised teststatistics for the null that the p -th eigenvalue diverges, based directly onthe estimated eigenvalue. The tests only requires minimal assumptions on thedata, and no restrictions on the relative rates of divergence of N andT are imposed. Monte Carlo evidence shows that our procedure has verygood nite sample properties, clearly dominating competing approaches when nocommon factors are present. We illustrate our methodology through anapplication to US bond yields with different maturities observed over the last30 years. A common linear trend and two common stochastic trends are found andidentied as the classical level, slope and curvature factors