원/달러 시장의 국면전환에 따른 국민연금기금 외환리스크 관리에 관한 연구

다국어 초록 (Multilingual Abstract)

Recently, in the world economy, the frequency of large and small crisis situations is increasing and the cycle is shortening. Accordingly, the risk of the foreign exchange market is expanding and is acting as a major factor in determining the profits and losses of overseas investments. Therefore, in order to manage foreign exchange risks that generate exchange gains and losses on foreign currency assets and foreign currency liabilities, the exchange rate time series distribution according to trends and transitions in the foreign exchange market must be accurately identified, and the exchange rate gains and losses that may arise through systematic risk monitoring and analysis based on this. We must accurately and objectively understand the scale and establish a foundation for establishing a wise response strategy. This study collected long-term time series of the won/dollar exchange rate, analyzed it through an objective quantitative model, monitored market transitions, and divided the regimes in a statistically significant manner. In addition, we estimated the foreign exchange gains and losses due to foreign exchange risks faced by market participants in each market phase as VaR, and based on this, we attempted to present a systematic risk management strategy. The main research results are as follows. First, the won/dollar exchange rate time series was derived from five breakpoints: September 29, 2001, June 25, 2005, July 19, 2009, July 19, 2014, and April 13, 2019. and the won/dollar foreign exchange market was divided into six market regimes. Second, as a result of comparing and analyzing the moments by regime, the regimes in which it was good to pursue foreign exchange gains were organized into regimes 3, 5, and 6, and the regimes in which there was a high possibility of incurring significant foreign exchange losses were organized into regimes 1, 2, and 4. The regimes in which extreme foreign exchange gains were possible were regimes 1 and 3, and the regimes in which foreign exchange losses were possible were regimes 2 and 5. Third, overall, the values obtained by the parametric method assuming a normal distribution underestimate the expected foreign exchange gains, and this bias in estimating the value of foreign exchange risk was further expanded as the level of confidence increased. On the other hand, under the 99% confidence level, the expected loss was overestimated in regimes 1 and 3 by the parametric method, and especially in regime 3, this result occurred in all confidence intervals due to the parametric method. The above results warn of the danger of assuming normal distribution using the parametric method when calculating VaR, even though the time series is not normally distributed. At the same time, special care must be taken if the skewness(S) of the FX time series has a large gap with the normal distribution(0), such as in regime 3(2.07) and regime 1(1.60).