Title: Sieve-based test for cointegration
Authors: Jan Andrew Reforsado - University of the Philippines Los Banos (Philippines) [presenting]
Erniel Barrios - University of the Philippines (Philippines)
Joseph Ryan Lansangan - University of the Philippines (Philippines)
Abstract: Cointegration testing is an important aspect of modeling in nonstationary time series data to avoid the possibility of observing spurious relationship among variables. Existing tests for cointegration usually exhibit less optimal behavior specially for short time series data. A vector error correction model is estimated through the backfitting algorithm, the fitted model is used in replicating the data through sieve bootstrap. The empirical distribution of eigenvalues from the lagged error correction matrix generated from the data replicates are used in testing for cointegration. Simulation study shows that the proposed nonparametric test yields size and power at least comparable to some well-known tests for cointegration.