Forecasting Financial Crises and Contagion in Asia using Dynamic Factor Analysis



We consider the external borrowing of the private sector (banks and non banks) and of the
public sector of each country from developed countries banks. In order to complete the
dataset describing thoroughly the external banking debt of the countries under investiga-
tion, we also include undisbursed credit commitments and local currency claims on local
residents. Furthermore, we include data on international bonds and notes issued by the five
Asian emerging economies under investigation.

We also include the money supply aggregate M2 (obtained from the International Finan-
cial Statistics, IFS, database of the IMF) of each country, and we convert each aggregate
into US dollars using the nominal exchange rate of the country versus the US dollar. Each
money based indicator of reserves provides a measure of the potential for resident-based
capital flight from the currency, since it is argued that, an unstable demand for money or
the presence of a weak banking system indicates a greater probability of such capital flight.
We also consider the total amount of imports (measured in millions of US dollars) of each
of the five countries under investigation.

Each of the aforementioned variables (in US dollars) is deflated by the country specific
stock of foreign exchange reserves (minus gold) in millions of US dollars in order to obtain
indicators of vulnerability. The data for the components of the
EMP index are obtained
from the International Financial Statistics (IFS) of the IMF database. As suggested by
Girton and Roper (1977), the measure of the
EMP index consists of a weighted sum of the
exchange rate depreciation rate (measured as unit of domestic currency per US dollar), and
of the change in the stock of US dollar denominated official reserves (minus gold) scaled by
the previous period base money (converted in US dollars). The weights are given by the
inverse of the corresponding sample standard deviations.

Finally, the EM P index of each country is also included in the dataset to account for the
role played by foreign currency mismatches in predicting a crisis event. This gives a total of
114 variable constituents for the dataset under investigation.

4.2 Empirical Results

The out-of-sample probability forecast are obtained through either recursive OLS or, in the
case of Probit modelling, recursive
ML estimation. First, we do not report the forecasting
results associated with the models given by either a naive predictor, described in eq. (
??),or
by the competing models described in equations (
??), (??), given that the corresponding KS

13



More intriguing information

1. Implementation of a 3GPP LTE Turbo Decoder Accelerator on GPU
2. CGE modelling of the resources boom in Indonesia and Australia using TERM
3. The name is absent
4. Foreign direct investment in the Indian telecommunications sector
5. The Impact of Individual Investment Behavior for Retirement Welfare: Evidence from the United States and Germany
6. The name is absent
7. Prizes and Patents: Using Market Signals to Provide Incentives for Innovations
8. Tax Increment Financing for Optimal Open Space Preservation: an Economic Inquiry
9. Determinants of U.S. Textile and Apparel Import Trade
10. Individual tradable permit market and traffic congestion: An experimental study
11. PEER-REVIEWED FINAL EDITED VERSION OF ARTICLE PRIOR TO PUBLICATION
12. The name is absent
13. The Role of State Trading Enterprises and Their Impact on Agricultural Development and Economic Growth in Developing Countries
14. Trade Liberalization, Firm Performance and Labour Market Outcomes in the Developing World: What Can We Learn from Micro-LevelData?
15. Cyber-pharmacies and emerging concerns on marketing drugs Online
16. NATURAL RESOURCE SUPPLY CONSTRAINTS AND REGIONAL ECONOMIC ANALYSIS: A COMPUTABLE GENERAL EQUILIBRIUM APPROACH
17. The value-added of primary schools: what is it really measuring?
18. THE ANDEAN PRICE BAND SYSTEM: EFFECTS ON PRICES, PROTECTION AND PRODUCER WELFARE
19. The name is absent
20. ROBUST CLASSIFICATION WITH CONTEXT-SENSITIVE FEATURES