Abstract
Abstract The decisions to reduce, leave unchanged, or increase a choice variable (such as policy interest rates) are often characterized by abundant status quo outcomes that can be generated by different processes. The decreases and increases may also be driven by distinct decision-making paths. Neither conventional nor zero-inflated models for ordinal responses adequately address these issues. This paper develops a flexible endogenously switching model with three latent regimes, which create separate processes for interest rate hikes and cuts and overlap at a no-change outcome, generating three different types of status quo decisions. The model is not only favored by statistical tests but also produces economically more meaningful inference with respect to the existing models, which deliver biased estimates in the simulations.
Highlights
The decisions to reduce, leave unchanged, or increase a choice variable are often characterized by abundant status quo outcomes that can be generated by different processes
This paper develops a flexible dynamic cross-nested ordered probit (CronOP) model that accommodates the unobserved heterogeneity of data-generating process by assuming three implicit decisions, and illustrates the model in the context of discrete adjustments to policy interest rates
The simulations demonstrate that the marginal effect (ME) estimates in the OP and nested ordered probit (NOP) models are biased when the underlying dgp is characterized by three types of zeros, and that the CronOP and CronCOP estimates systematically provide superior coverage probability (CP) as well as smaller bias
Summary
Ordinal responses, when decision-makers face a choice to reduce, leave unchanged or increase a price (consumption, rating, or policy interest rate) are often characterized by abundant no-change outcomes that may emerge from fundamentally different behavioral mechanisms. If the stance is loose/tight, the policymakers can cut/hike the rate by a certain discrete amount or leave it unchanged These unidirectional amount decisions, which are conditional on the regime, determine the magnitude of the rate adjustment, intensifying or weakening the policy inclinations. The model simultaneously estimates the three dynamic (with the lags of the observed policy choices among regressors) OP equations, which represent the latent regime and amount decisions, and allows for a possible correlation among them Using this interpretation, we can classify zeros into three types and describe how they arise: the “always” or “neutral” zeros, which are directly generated by a neutral reaction to economic conditions, and the two types of “not-always” zeros – the “loose” and “tight” zeros – which are generated by loose or tight policy inclinations and are offset by tactical and institutional reasons. If a certain variable has an impact on both latent decisions, the OP model cannot reveal the distinct effects on the probabilities of different types of zeros (with respect to both a sign and a magnitude), incorrectly estimates that variable’s total impact by focusing on the observed zeros, and produces the misleading estimates of the choice probabilities and marginal effects
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
