시장경제연구(市場經濟硏究) : 금융부문에서의 신용리스크와 시스템적 리스크의 측정

최경욱 ( Kyong Wook Choi ),형남원 ( Nam Won Hyung ),정원경 ( Won Kyung Jung ) 서강대학교 시장경제연구소(구 서강대학교 경제연구소) 2015 시장경제연구 Vol.44 No.2

본 연구에서는 신용리스크 측정의 대안적 방법을 제시하고, 이러한 신용리스크의 금융기관 간 연계효과의 측정을 통한 시스템적 리스크의 추이를 파악했다. 신용리스크는 극치이론에 기초해 자산의 시장가치가 부채를 이용하여 설정한 임계점보다 작아질 확률로 정의한 건전성지표로 추정하였다. 분석의 대상으로 2002년부터 2013년까지의 은행권역과 비은행권역을 선정하고, 비은행권역으로 손해보험, 증권, 상호저축, 여신전문 등의 업종을 선택하고, 그리고 동일 업종 내에서는 3~4개의 기업을 선정하였다. 실증분석에서 개별 기업의 신용리스크를 측정하고, 동일 업종 내에서 각 기업 간의 신용리스크의 전이를 분석하였다. 그리고 업종 간의 전이, 특히 은행과 비은행 사이의 전이도 분석하였다. 실증분석을 통해 향후 국내에서 금융업 내부에서 발생하는 내생 위기뿐만 아니라 해외부분에서 발생하는 불안정 요소의 영향이 신용리스크 발생 및 전이에 중요한 역할을 하는 것으로 나타났다. In this study, we suggest an alternative method of measuring credit risk and investigate systemic risk by analyzing the linkages among financial institutions. By using extreme value theory, we also build a prudential indicator. Our sample period covers 2002 to 2013. We measure information from the variation in the fundamental value of the stock price of banking and non-banking firms in the Korean stock market. Additionally, we compare the financial credit risk taken from the debt level of banking and non-banking institutions as well as the credit risk of the financial firms analyzed in this study. We conclude that the credit risk of Korean financial institutions is not only affected by domestic crises but also by foreign financial market crises.

환율변동성과 국제무역 관계 분석

최경욱 ( Kyong Wook Choi ) 慶熙大學校 社會科學硏究院 2010 社會科學硏究 Vol.36 No.1

본 연구에서는 대미환율변동성이 캐나다, 한국 그리고 멕시코에 미치는 영향을 최근에 개발된 계량기법을 이용하여 분석하였다. 대미환율변동성을 마코프변환을 따르는 분산을 통해 측정하였다. 또한 인과관계를 살펴보기 위하여 마코프변환 자기회귀모형을 이용하여 추정하였다. 연구결과 캐나다와 한국에서만 대미환율변동성이 통계적으로 의미 있는 음의 결과를 보이고 있음을 알 수 있다. In this study, we re-examine the effect of exchange rate volatility on the United States' bilateral real exports to Canada, South Korea, and Mexico using recent econometric techniques. In this effect, we use the exchange rate volatility as measured by the Markov switching-variance process. To identify the sign of the causal effects, we use the Markov Switching Vector Autocorrelation (MS-VAR) model. This model produces mixed results. Canada, and Korea provide statistically significant, negative relationship between exchange rate variability and trade but Mexico does not.

시장경제연구 : 자본이동과 환율간의 관계 분석

최경욱 ( Kyong Wook Choi ) 서강대학교 시장경제연구소(구 서강대학교 경제연구소) 2014 시장경제연구 Vol.43 No.3

본 연구에서는 먼저 환율과 자본이동 간의 동태적 관계를 살펴 본 결과 Granger 인과관계 분석 결과에서는 환율 변화율이 자본이동의 변화 예측에 설명력이 있음을 보였다. 충격반응함수 분석 결과는 환율 변화율 충격이 자본이동에 미치는 영향은 국제 금융위기 기간과 이후 사이에 상당히 다른 형태의 모습을 보이고 있다. 분산 분석 추정 결과도 금융위기 기간 동안 환율 변화율 변동 설명력이 금융위기 이후 보다 크게 나타나고 있다. 마지막으로 다변량 GARCH-in-Mean 모형을 이용해서 환율 변화율과 자본이동의 변동성이 각 변수에 미치는 영향을 보면 환율변동성이 국제 자본이동에 미치는 효과가 통계적으로 유의하게 나타나고 있다. 따라서 정책적인 측면에서 급격한 국제 자본이동의 문제점을 제어하기 위해서는 환율변동성을 적절한 수준으로 조정해야 할 것이다. This study investigates the dynamic relationship between the Won/USD exchange rate and capital flows by adopting the VAR model. Key findings from the study are the following. First, the Granger causality test shows that the changes in the Won/USD exchange rate provide statistically significant predictability to capital flows. Second, the impulse responses of capital flows to the changes in the Won/USD exchange rate exhibit stronger effects during the global financial crisis period than after the global financial crisis. Similarly, variance decomposition results show similar results. Finally, the multivariate GARCH-in-Mean model shows that the volatility of the change in the Won/USD exchange rate has statistically significant effects on capital flows.

패널 자료를 이용한 지역별 주택매매가격 분석:

구희일(Hee Il Koo),최경욱(Kyong Wook Choi) 한국주택학회 2018 주택연구 Vol.26 No.1

본 연구는 패널 VAR 모형으로 글로벌 금융위기 전후 기간의 지역별 주택매매가격 변화를 기간별로 비교 분석하였다. 주택매매가격에 영향을 주는 변수로는 전입건수와 주택거래실적, 소매판매액지수, 가계대출 잔액을 이용하였다. 분석 결과, 전체 기간 중 전국 기준 주택매매가격은 주택매매가격 자체와 주택거래실적에 유의한 양(+)의 반응을 보였다. 글로벌 금융위기로 기간을 구분한 분석에서는 전국 기준주택매매가격이 위기 이전 기간과 위기 이후 기간에 각각 소매판매액 충격과 가계대출 충격에 유의한 양(+)의 반응을 보였다. 수도권 지역의 주택매매가격은 위기 이전 기간에 소매판매액 충격에 양(+)의 반응을 보이다가 위기 이후 기간에는 음(-)의 반응을 보였다. 한편, 비수도권 지역은 위기 이후 기간에 가계대출 충격이 주택매매가격에 양(+)의 영향을 주는 것으로 나타났다. 이와 같은 결과들을 종합할 때, 향후 부동산관련 정책 수립 시 소비재와 투자재의 성격을 모두 갖고 있는 주택매매가격의 특성이 고려되어야 할 것이다. 아울러 시기와 지역에 맞는 다양한 변수들에 대한 종합적인 분석도 필요할 것으로 보인다. This study empirically examined the changes in the domestic housing market by region in the aftermath of the global financial crisis. The panel VAR model was used to estimate the change in housing sales price in the period before and after the global financial crisis, using several variables such as internal migration, housing transactions, retail sales index, and household loan balances. As a result of the analysis, nationwide housing sales price showed a positive response to housing sales price and transactions. At the time of the global financial crisis, nationwide housing sales price showed a positive response to the retail sales shock and the household loan shock in the pre-crisis period and post-crisis period respectively. In metropolitan areas, housing sales price showed a positive response to the retail sales impact in the pre-crisis period and a negative response in the post-crisis period. On the other hand, in non-metropolitan areas, the impact of the household loan shock on the housing sales price positively affected in the post-crisis period. The results of this study demonstrated that the housing sales price vary not only in time and region but also in the underlying variables that affect housing price. Therefore, characteristics of housing sales price, which include both consumer and investment properties, should be considered when establishing real estate related policies in the future.

국내외 금융시장의 연계성 변화 분석: 외환위기와 글로벌 금융위기 기간을 중심으로

유복근 ( Bok Keun Yu ),최경욱 ( Kyong Wook Choi ) 한국국제경제학회 2010 국제경제연구 Vol.16 No.1

금융연구 : 금융기관 신용리스크의 파급효과와 시스템 리스크 측정

김명석 ( Meong Suk Kim ),최경욱 ( Kyong Wook Choi ),형남원 ( Nam Won Hyung ) 한국금융연구원 2013 금융연구 Vol.27 No.2

We have experienced a tremendous global financial crisis from 2008 which threatened the soundness of the global financial system. In order to overcome the serious economic problems that have arisen, most developed countries have used quantitative easing policy along with expansionary fiscal policy. For example, FRB purchased $85 billion in bonds a month beginning September 2012, and has committed to maintain the benchmark interest rate at low levels until at least mid-2015. During this process, the fiscal deficit has been growing fast, dropping the US credit rating in the meanwhile. This growing fiscal deficit has also contributed to the European economic crisis. On the other hand, the Korean financial market has kept a solid fundamental economic system, which has helped Korea to recover from the financial crisis relatively quickly as compared to other developed countries. Despite this successful recovery, we still must take precautions against financial crises in the near future. The reason we attempt to predict financial crises is in order to minimize the losses of the financial system. To do so, we have to analyze the credit risk of each financial institution and also find potential system risks. In this study, we try to explain the spillover effect of credit risk between financial institutions. We develop the spillover index model to measure how financial institutions` credit risks affect each other. We extend Diebold and Yilmaz`s (2009) spillover index in order to use the VAR model to measure how specific sectors` variables affected one another. We also identify the magnitude of the system risk as the spillover index. We use 2 variables as credit risk: domestic banks` CDS (Credit Default Swap) spread and several banking and non-banking EDF (Expected Default Frequency). We obtain the CDS spread data from Markit. We have 11 banks` CDS spread data : the banks include Kookmin Bank (KB), Woori Bank (WR), Shinhan Bank (SH), Hana Bank (HN), Korea Exchange Bank (KE), Citibank Korea (CT), SC Korea First Bank (SC), Industrial Bank of Korea (IB) Korea Development Bank (KD), Nonghyup Bank (NH), and Export-Import Bank of Korea (EX). It is important to identify the magnitude of the relative change, period by period, rather than the absolute value of the change itself. Using the CDS spread, we found that prior to the financial crisis the spillover index was at a low level and that after the financial crisis, it has reached a high level. The EDF data was collected from the following : in the banking sector, we include 7 banks. In the non-banking sector we use 7 insurance companies, 7 S&L banks, and 5 lease companies. We can use EDF as supplementary data, as the CDS spread data is still more relatively important. CDS is decided by banks, while EDF is also controlled by other sectors, such as insurance companies, S&L, and lease companies, of which some of the S&L companies have been the cause of large financial turmoil in the Korean financial system in the past few years. In our study, we find that credit risk spillover is not that high in the banking sector. Additionally, the problem of some of the S&L companies` crisis is due to the problems of the S&L system itself; there is very little spillover between S&L and other financial sectors, such as the banking sector. In order to ensure the soundness of the financial system itself, we have to develop a strict macro-prudential policy. The limitation of this approach is that we provide how much the credit spillover increases but we do not suggest the cause of this credit spillover increase. We are going to propose the cause of this increase in future research.

혼합자료 샘플링 모형을 통한 GDP 예측

형남원 ( Nam Won Hyung ),최경욱 ( Kyong Wook Choi ),최병재 ( Byung Jae Choi ) 한국국제경제학회 2016 국제경제연구 Vol.22 No.2

경제전망을 위해 다양한 예측 방법을 활용한 GDP의 예측이 논의되어 왔다. 본 연구에서는 혼합주기 샘플링(MIDAS) 모형과 AR-MIDAS 모형을 이용한 GDP의 단기전망을 실시하였다. 예측을 개선하는 방법으로 제시되는 MIDAS 모형은 월별 거시경제변수를 활용하여 당분기, 1분기 혹은 2분기 후의 GDP 예측에 월별 자료의 정보가 유용함을 실증적으로 보여준다. 이는 MIDAS 모형이 다른 경쟁 모형에 비해 월별 자료가 갖는 풍부한 정보를 좀 더 효과적으로 예측에 이용하기 때문인 것으로 보인다. 개별 거시경제변수를 이용한 MIDAS의 예측은 선택된 변수에 따라 상대적 우열이 다르게 나타나지만, 단순평균, 가중평균 혹은 중위값 등의 방법에 의한 결합예측의 경우 예측력이 뚜렷하게 개선되는 것으로 나타났다. 본 연구를 통해 GDP 예측에서 월별 자료를 이용한 MIDAS 모형의 예측방법 및 이들 예측치의 정보를 종합하는 결합예측 방법의 사용은 GDP의 단기예측의 예측력을 크게 개선시키는 것으로 나타났다. Many methods have been developed to forecast GDP. In this study, we use the MIDAS and AR-MIDAS models to forecast GDP at short forecast horizon. We find that the MIDAS model uses monthly macroeconomic variables more efficiently compared to other models for nowcasting and forecasting quarterly GDP growth. The MIDAS model contends for superiority in forecasting GDP when using monthly macroeconomic variables. However, when we utilize combined forecast by the simple average, weighted average or median methods, we find that most of the combinations perform well relative to the benchmark. In conclusion, we find that the pooling of MIDAS or AR-MIDAS perform better in terms of nowcasting and forecasting quarterly GDP growth.

교육환경과 아파트 전세가격간의 관계 분석

윤병우(Youn, Byung Woo),최경욱(Choi, Kyong Wook) 한국부동산학회 2011 不動産學報 Vol.47 No.-

1. CONTENTS (1) RESEARCH OBJECTIVES This study investigates the effect of the strong educational environment on the rental price of Kangnam areas, including Kangnam-Gu, Seocho-Gu, and Songpa-Gu. The apartment rental market differs from the apartment purchasing market because there is no investment purpose involved. In this sense, we identified the variable factor on the price of residence in Kangnam to be the effect of its educational environment. We used the fundamental real estate variables as structural characteristic variables. We used five different educational variables to analyze the effect of educational environment effects. (2) RESEARCH METHOD Empirical analysis’ Hedonic Price Model, Hierarchial Linear Model. (3) RESEARCH RESULTS From two models, we found that the majority of the variation of apartment rental price market depends on the regional educational environment effect. For example, in the Hedonic Price Model, we used five educational environment variables, such as the distance from middle schools, the number of cram schools, the ratio of college bound students to high school graduates in that area, the number of students who gained entrance to Seoul National University from that area, and the ratio of students transferring to that area from other areas. Interestingly, all five variables are statistically significant. On the other hand, the more sophisticated Hierarchial Linear Model suggests that only four variables are statistically significant, and that the number of cram schools is not statistically significant. 2. RESULTS Our research suggests very interesting real estate and educational policies. The empirical results suggest that all five variables have important implications, but because of the development of the online education market, the importance of the number of cram schools on the real estate rental market is decreasing. People relieve that one of the most important factors to Educational success is the educational environment in which they are raised in, and the real estate rental market reflects this. When we develop new real estate policy, we need to take into account the implications of this study.