Estimation of the Heteroskedastic Canonical Contagion Model with Instrumental Variables

PLoS One. 2016 Dec 28;11(12):e0168967. doi: 10.1371/journal.pone.0168967. eCollection 2016.

Abstract

Knowledge of contagion among economies is a relevant issue in economics. The canonical model of contagion is an alternative in this case. Given the existence of endogenous variables in the model, instrumental variables can be used to decrease the bias of the OLS estimator. In the presence of heteroskedastic disturbances this paper proposes the use of conditional volatilities as instruments. Simulation is used to show that the homoscedastic and heteroskedastic estimators which use them as instruments have small bias. These estimators are preferable in comparison with the OLS estimator and their asymptotic distribution can be used to construct confidence intervals.

MeSH terms

  • Computer Simulation
  • Data Interpretation, Statistical*
  • Humans
  • Models, Economic*
  • Models, Statistical*

Grant support

This work was supported by São Paulo Research Foundation (FAPESP) 2013/00506-1 and 2013/22930-0, CNPq, Capes.