NEW TIME SERIES COMMANDS

Time series with some or all of the parameters transitioning among a finite set of unobserved states with unknown transition points are estimated using Markov-switching models. Applications of Markov-switching models are the growth rate of GDP, monthly stock returns and transition between two different states (specific for social and health sciences). Users can now fit Markov-switching models with the new command mswitch which provides two ways of modeling the switching process: autoregressive (AR) which better fits processes that change slowly and dynamic regression (DR) which better fits processes that transition rapidly.

Stata 14 also allows the ability to obtain tables of transition probabilities and expected state durations using two new postestimation commands along with the ability to test for structural breaks with single unknown date, single known date and multiple known dates.

A selected list of available models and features in Stata 14 for Time Series:

Available Models	Available Features
v-switching dynamic regression	Test for structural breaks: single unknown date, single known date and multiple known date
Markov-switching autoregression	Transition probabilities
	Expected state durations

Return to Stata 14 New Features