Bowen Fu, "Bubbles and Crises: Replicating the Anundsen et al. (2016)
Results", Journal of Applied Econometrics, Vol. 34, No. 5, 2019,
pp. 822-826.

For narrow replication, I use the dataset of Anundsen et al. (2016) that 
is available on the JAE data archive. See

  http://qed.econ.queensu.ca/jae/datasets/anundsen002/

This dataset covers 20 countries and a period from 1976 to 2014. For 
wide replication, I use almost the same dataset, but I do not remove the 
observations during or immediately after a financial crisis, and I drop 
observations before 1981.

The dependent variable is a dummy variable for the pre-crisis state.
The independent variables include private credit growth, household
credit-to-GDP gap, NFE credit-to-GDP gap, house price-to-income gap,
output gap, global credit/GDP gap, global house price to income gap,
exuberance house price to income gap, exuberance credit/GDP, non-core
funding gap, and equity ratio. 

All data files are ASCII files in DOS format. They are zipped in
bf-data.zip. Unix/Linux users should use "unzip -a".


Data for Narrow replication:

replication_data_model1 is for constant parameter logit model 1 and
has 1990 observations.

replication_data_model2 is for constant parameter logit model 2 and
has 1843 observations.

replication_data_model3 is for constant parameter logit model 3 and
has 1333 observations.

replication_data_model4 is for constant parameter logit model 4 and
has 926 observations.


Data for Wide replication:

fullpanel1.csv is for TVP M1 has 1877 observations.

fullpanel2.csv is for TVP M2 has 1877 observations.

fullpanel3.csv is for TVP M3 has 1723 observations.

fullpanel4.csv is for TVP M4 has 913 observations.