The persistence of Taiwan's output fluctuations: an empirical study using innovation regime-switching model

In this article we examine the persistence nature of Taiwan's aggregate output fluctuations by using the 'innovation regime-switching' (IRS) model in which the effect of an innovation may be permanent or transitory, depending on an unobservable state variable that follows a first order Markov chain. By applying the IRS model to Taiwan's real GDP data, we find that during the 1961 to 2000 period 61% (39%) of the real output shocks are likely to have permanent (transitory) effects. Moreover, the innovations in the officially identified expansion (contraction) are more likely to have a permanent (transitory) effect. These results are similar to those found in many studies of US real output fluctuations, e.g. Beaudry and Koop (1993), Kim and Nelson (1999) and Kuan et al. (2005). However, we also find that Taiwan's output dynamics have changed drastically ever since year 2000. In particular, the shocks to real GDP have become more likely to have only transitory effect, even during the period of post-2001:IV expansion.