Identifying Multiple Regimes in the Model of Credit to Households

Abstract

This research proposes a new method to identify the differing states of the market with respect to lending to households. We use an econometric multi-regime regression model where each regime is associated with a different economic state of the credit market (i.e. a normal regime or a boom regime). The credit market alternates between regimes when some specific variable increases above or falls below the estimated threshold level. A new method for estimating multi-regime threshold regression models for dynamic panel data is also demonstrated.

Keywords: credit boom, threshold regression, dynamic panel

JEL Classification: E51, C23, C51