Towards a new early warning system of financial crises

Abstract

Standard-Nutzungsbedingungen:Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden.Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen.Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte.

Patterns of Current Account Adjustment—Insights from Past Experience

“…As an alternative, we also estimate specifications that assign a separate value for the dependent variable in the two years after adjustment, as in Bussière and Fratzscher (2006). The results (not reported here) are broadly similar.…”

“…In our specification, we assign a non-zero value to our independent variable not only in the exact quarter where the adjustment starts, but also in the eight quarters before. 17 The approach is appealing from a policy viewpoint, as it allows to signal adjustments not just in the current quarter but over a horizon of two years, and from an econometric viewpoint, as it allows to avoid the use of lagged independent variables (see Bussière and Fratzscher (2006) for an application of a similar technique in a context of early warning systems for currency crises). The observations immediately after the start of adjustment (2 years) are excluded from our estimations so as to avoid any potential bias that may arise when the model does not distinguish post-adjustment times from tranquil times.…”

Bubbles and Crises: The Role of House Prices and Credit

“…This is particularly relevant in the context of the CCB due to the 12 month implementation lag. 2 For this reason, we follow Bussiere and Fratzscher (2006) and define our dependent variable, Y i,t , as a forward-looking variable…”

“…This is done to avoid the post-crisis bias, as discussed in e.g. Bussiere and Fratzscher (2006). 3 Given the definition of our dependent variable in (1), and considering a logit specification, the probability of a financial crisis in country i over the specified horizon is given by p i = P r (Y i,t = 1) = Φ (α i + β x i,t ) = exp(α i + β x i,t ) 1 + exp(α i + β x i,t )…”

Current Account Reversals in Industrial Countries: does the Exchange Rate Regime Matter?

“…This approach is used by Bussiere and Fratzscher (2006) in the context of early warning models for predicting financial crisis. The strategy increases the model's capacity to identify statistically significant determinants in the regressions and allows attenuating potential endogeneity concerns.…”