Cointegration between censored variables

Abstract

The study investigates the cointegration between nonstationary censored endogenous variables in the case of restrictions. It examines that if common trend (cointegration) can be detected between the time series, then what changes need to occur to eliminate the error correction mechanism between the processes. A not rare anomaly is investigated; the case of one of the time series' value becomes zero due to an unexpected event.

After reviewing the concept of random walk and cointegration, the paper mentions the Tobit cointegrated vector autoregressive model. It presents an error correction model between censored variables belonging to this model family.