Forecasting Financial Crises and Contagion in Asia using Dynamic Factor Analysis

Note: The numbers corresponding the row labelled ARDL_pc are indicators of probability forecast
performance (of a crisis event defined as 1.5 standard deviation above the mean) when using the ARDL
model, with the principal components as exogenous regressors. The numbers corresponding the row labelled
Probit_pc are indicators of probability forecast performance when using the Probit model (estimated by ML),
with the principal components as exogenous regressors. In each cell, the number preceding the semicolon
is the KS score, whereas the number right after the semicolon in each cell is the Matthews correlation
coefficient. NA stands for Not Available and it is due to values both equal to zero for the numerator and
the denominator of theMatthews correlation coefficient.

Table 2: Probability Forecasts of a crisis event (2 std)

Note: See Note to Table 1. The crisis event is defined as two standard deviation above the mean

Note: In this Table we report indicators of probability forecast performance (the KS score and the
Matthews correlation coefficient on the l.h.s. and r.h.s of each cell, respectively) associated with the ARDL
model which uses the principal components as exogenous regressors. The principal components have been
extracted from a dataset without the ratio of total external debt to the stock of international reserves in
each country.

Note: see Note to Table 3. The principal components have been extracted from a dataset without the
ratio of short term external debt to the stock of international reserves in each country

Note: see Note to Table 3. The principal components have been extracted from a dataset without the
ratio of long term external debt to the stock of international reserves in each country

20

<<   <   18   19   20   21   22   23   >   >>