Bulletin Spring‧Summer 2004

denotes a state of strongest resilience while a score of '1' denotes least resilience. The five signals can generate 32 possible combinations of the 'strong' and 'weak' signals. Each combination will be given a resilience score presented in a decision matrix. For instance, a combination of 5 strong signals will be given a resilience score of 5 while a combination of 5 weak signals will be given a resilience score of 1 and so on. In the construction of this decision matrix, expert opinions from economists on the relative importance of various indicators are incorporated to assess the resilience level of the sector. The aggregated resilience score is then computed by taking a probability-weighted average of resilience scores given to each indicator. The probability is obtained from the artificial intelligence engine used in the preceding step. Next, the classification and regression tree (CART) approach is used to develop a series of decision rules through which resilience scores will be assigned to the observation, in such a way that the number of observations falling into the five score groups of '1' to '5' will match as much as possible the number of observations originally classified into the groups in the previous step. The product of the CART process is a decision tree, which will be used for assigning ratings to future observations. Some key findings are summarized in the box below. Some Key Indicators of Resilience • ‘Short-term external debt cover' emerged as the most important indicator of external sector resilience as reflected by the number of times it appeared on the tree. This indicator is crucial as it reflects an economy's ability to repay short-term external obligations, especially in times of financial shock. • The ‘net international investment position (IIP)' is the next most important factor in assessing the resilience of the external sector, as it is an indication of an economy's stock of wealth, or the size of cushion available in times of financial crisis. Wh i le a negative IIP does not necessarily suggest high vulnerability, a strong IIP w i ll suggest a large cushion to absorb financial shocks. • ‘Export growth' is also a significant factor in determining external sector resilience as reflected by the indicator appearing at the top of the decision tree. It provides an indication of income f l ow i ng into an economy. international finance centre Research News 4 5