|
Abstract:
|
We consider estimation of structural breaks in vector autoregressive models. In practice, the number of change points is usually assumed to be known and small, because a large would involve a huge amount of computational burden for parameters estimation. By reformulating the problem in a high-dimensional variable selection context, the group orthogonal greedy algorithm is proposed to estimate the structural breaks in the model. Desirable theoretical results about consistency and convergent rate of the break-location estimates are derived to lend support to the proposed methodology.
|
