Sub-Theme 3 | Policy Recommendations
While a statistical early warning model is a useful device for assessing the vulnerability of an economy, it is also clear that no quantitative model is perfect. The deficiencies of quantitative early warning models imply that qualitative monitoring is important to preventing a crisis. Qualitative monitoring helps identify imminent risk and thus supplements the early warning models. Indeed, quantitative models and qualitative monitoring are two pillars of an early warning system.
It is recommended, therefore, that a government in an emerging market country strengthen qualitative monitoring of financial markets, financial institutions, and foreign markets in tandem with developing quantitative models. Given the limit of quantitative models, qualitative monitoring is especially crucial in emerging market countries where high quality data is not available as compared to advanced countries. Qualitative monitoring supplements the numerical outcomes of quantitative statistical models during the decision phase regarding the risk of crisis and subsequent policy responses.
For conducting qualitative monitoring and operating quantitative models, the government needs to establish an organization. The organization can be an independent institute or a department in either central bank or government. In Korea, the Korea Center for International Finance (KCIF) was established in April, 1999 by the Korean government and the Bank of Korea. Having learned that the Asian financial crisis is caused by foreign currency liquidity problem, the Korean government needed a public institution that could assist the government in preventing currency crises. Thus, the main responsibilities of the KCIF are monitoring international financial markets and operating early warning models.
Other recommendations are provided for the improved development and operation of an early warning system. First, effective operation of an early warning system requires the government to prepare a crisis management manual. In the crisis management manual, contingency policy directions are described based on the degree of warning signals from the early warning system.
Second, it should be noted that any statistical early warning model is developed based on past crisis episodes. For example, the signal approach used in the EWS for currency crisis presumes that abnormal behaviors of leading indicators on the eve of past crises would be repeated in the future. However, the underlying vulnerabilities and triggering events could be different in future crises. Therefore, in principle, there is no guarantee that a model that performed well in the past would continue to work well in the future.
In particular, emerging market economies may experience rapid changes in the structure of their own economy. If an early warning model fails to incorporate these changes properly and in a timely manner, the model would perform poorly with more missed calls and false signals. For example, although large current account deficits certainly represent vulnerability, volatile capital flows across borders have become increasingly more important in recent years. Therefore, new leading indicators predicting the capital and financial account balance should be added to the model to signal a crisis caused by sudden capital outflows. In addition, as longer time series data become available, noise-to-signal ratios, threshold values, and other statistics should be re-estimated, leading to changes in the structure of composite index. Indeed, continuous updates and revisions are critical to improving the performance of early warning models.
Third, ensuring the quality of leading indicators is essential to a successful early warning system. To improve overall statistical capacity, the central statistical agency should collect, process, and disseminate data in accordance with international standards. The statistical agency needs to enhance the expertise of its staff members through various training programs in collaboration with international organizations. It is also important for the agency to adapt new methodologies and technologies and to upgrade the IT infrastructure.
Fourth, it is recommended that emerging market countries develop a simple model in the early stage and then evolve into more sophisticated models over time. It should also be noted that early warning signals based on a single quantitative model are subject to model risk. It is always possible for the model to miss a warning call or send a false alarm. In response to these possibilities, therefore, it is recommended that auxiliary models be developed as supplementary tools. The main model and auxiliary models may use different statistical methodologies and data
|Subject||Economy < Financial Policy|