An Econometric Analysis of Some Models for Constructed Binary Time Series

Abstract

Macroeconometric and financial researchers often use binary data constructed in a way that creates serial dependence. We show that this dependence can be allowed for if the binary states are treated as Markov processes. In addition, the methods of construction ensure that certain sequences are never observed in the constructed data. Together these features make it difficult to utilize static and dynamic Probit models. We develop modeling methods that respect the Markov-process nature of constructed binary data and explicitly deals with censoring constraints. An application is provided that investigates the relation between the business cycle and the yield spread.