Earticle

This study investigates the relationships between three leading cryptocurrencies— Bitcoin, Ethereum, and Ripple—and six major implied volatility indices as proxies for stock market uncertainty: VIX, VXD, VSTOXX, VDAX, VXEFA, and VXEEM. We employ nonparametric causality-in-quantiles and quantile-on-quantile approaches to examine nonlinear causal effects and dependence structures across various cryptocurrency market states and levels of uncertainty. Our findings reveal a one-way information flow from cryptocurrency returns to stock market uncertainty, with stronger predictive power during periods of low uncertainty. This suggests that while stock market uncertainty may not reliably predict cryptocurrency returns, cryptocurrency-related information can significantly influence stock markets, particularly during low uncertainty, likely driven by investor attention to cryptocurrencies for diversification or speculative purposes. The quantile-on-quantile analysis shows that changes in implied volatility generally have a negative impact on cryptocurrency returns, with these effects being more pronounced at lower quantiles. Furthermore, cryptocurrencies’potential to act as hedges or safe havens against stock market uncertainty emerged only under extremely bullish market conditions and has significantly diminished over time due to the increasing integration of cryptocurrency and stock markets.

본 연구에서는 그래프 신경망 모형을 포함한 6개의 기계학습모형을 활용하여 한국 기업의 신용변동을 예측했다. 이를 예측하기 위해 수익성, 성장성, 안정성, 활동성 지표들을 활용하였고, 주성분 분석을 진행하여 지표별로 두 개씩 활용했다. 또한, 신용등급변동이 없는 경우가 대다수인 불균형 데이터를 처리하기 위해 SMOTE 방법 론을 활용하여 데이터를 오버샘플링했다. 본 연구에서는 산업구조를 반영하기 위해 이를 나타내는 상장시장, 데이터 시점, 현재 신용등급과 업종의 네 분류로 나누어 모형을 만들고 이를 최종적으로 결합하여 예측 결과를 도출했다. 결과를 살펴보면, 모든 모형이 데이터 비중에 맞게 선택하는 벤치마크 모형보다 우수하고, 그래프 신경 망 모형이 특히 더 우수한 성능을 나타내는 것을 확인할 수 있다. 또한, 변수 중요도를 분석한 결과, 성장성 지표와 활동성 지표가 중요한 것을 확인할 수 있었고, 그래프 신경망에서는 업종 그래프와 상장시장 그래프가 성능이 좋은 것을 확인할 수 있었다.

Credit rating is an essential component of the financial sector, as it evaluates a firm’s capacity to meet debt obligations. In the Republic of Korea, rating agencies assign levels ranging from AAA to D, with additional modifiers, and these ratings significantly affect financing conditions. Traditional methods typically rely on logistic regression and selected financial variables, yet these approaches often face difficulties in capturing the intricate or nonlinear patterns present in corporate financial data. In response to these challenges, researchers have increasingly turned to advanced machine learning algorithms that can account for more complex relationships. Nevertheless, their deployment is limited by relatively small datasets—particularly among smaller firms—and by concerns regarding model interpretability. The present study proposes a machine learning framework, including a Graph Neural Network (GNN), to predict rating changes in Korean firms. The data set spans 2010 to mid-2024 and includes 182 firms, yielding 1,417 year-level samples. Each annual observation is labeled according to whether its credit rating was upgraded (+1), downgraded (–1), or left unchanged (0). Because most observations lie in the unchanged category, the data are highly imbalanced. To address this imbalance, the Synthetic Minority Over-sampling Technique (SMOTE) is employed, generating additional samples in the minority classes. In addition, Principal Component Analysis (PCA) is utilized to reduce the dimensionality of sixteen indicators representing profitability, growth, stability, and activity. Six models are assessed: logistic regression, LASSO, random forest, support vector machine, Light Gradient Boosting Machine (LGBM), and a GNN. The GNN-based approach is noteworthy for modeling financial indicators or corporate attributes as nodes within a graph, with edges delineating the relationships among these nodes. Such a representation allows for the capture of latent dependencies that are difficult to detect in methods that treat predictors independently. Furthermore, each corporate sample belongs to discrete subgroups determined by listing market (KOSPI or KOSDAQ), data period, rating band, and industry classification. The proposed hierarchical GNN merges these subgroup-specific networks to produce a consolidated prediction, thereby incorporating both firm-level attributes and group-level characteristics. When compared to a benchmark that classifies samples randomly in proportion to the observed class distribution, all six machine learning algorithms demonstrate superior performance. The GNN shows the highest precision and F1-scores, suggesting that it is particularly effective at identifying upgrades and downgrades, which are far less common than no rating changes. Nonetheless, like other models, it finds rare rating shifts more challenging to predict, highlighting the impact of data imbalance and the difficulty of forecasting uncommon events. An inspection of feature importance across models underscores the significance of growth and activity metrics, implying that sales expansion, equity growth, and the efficient use of assets offer robust signals of rating volatility. Moreover, the GNN indicates that distinguishing firms by industry group is especially influential, possibly because each sector’s distinctive regulatory, economic, and financial traits shape its credit risk profile. Compared to certain deep neural networks that demand extensive datasets, the GNN-based method presented here is relatively more practical in settings with limited data, including smaller firms with incomplete rating histories. Additionally, this approach provides improved transparency, as the graph architecture clarifies how different financial indicators or subgroups collectively affect rating transitions. Future work may benefit from enlarging the dataset, experimenting with alternative oversampling strategies such as ADASYN, and examining cost-sensitive learning to mitigate the imbalance problem further. Investigations might also consider alternative graph structures that connect entire firms as nodes and delineate inter-firm relationships or incorporate advanced architectures such as Transformers or LSTM networks. In summary, the findings suggest that a GNN-based framework can improve credit rating predictions by capturing complex interactions that traditional or other advanced machine learning methods may overlook. While data imbalance is still a problem, its consequences are somewhat mitigated by SMOTE. The significance of growth, activity, and sector-specific characteristics suggests that more accurate and comprehensible rating projections can be produced by integrating richer and more interconnected data. In the end, more investigation and more extensive data gathering should improve the precision and dependability of credit rating systems, leading to a better comprehension of the dynamics of corporate finance.

KRX 유가증권시장은 2019년 4월 29일 시가단일가매매 호가접수시간(“시가단일가매매시간”)을 1시간에서 30분으로 단축하였다. 이 논문은 시가단일가매매시간에서 외국 인과 기관투자자가 가격발견에 미치는 영향력을 파악하고자, 이 사건을 전후로 시가단 일가매매시간을 5분 단위로 나누어 투자자별 호가 제출 패턴과 WPC(“가중가격공헌 도”)의 횡단면 결정요인을 분석하였다. 사건 전후 50 거래일 동안 모든 호가를 분석한 결과 첫째, 외국인과 기관투자자는 개인투자자와는 달리 장 후반에 호가를 집중·제출 하는 패턴을 보인다. 또한 같은 시간대에 정정·취소호가 비중도 매우 높아 시가단일 가매매 체결시점에 가까워질수록 가격을 적극 조정하며 정보거래자로서 시가 발견에 큰 영향을 끼친다. 둘째, 시가단일가매매에서 WPC를 결정하는 요인은 장 개시 초반 (“WPC 최초 계산 구간”)과 장 후반(“WPC 조정 구간”)이 서로 반대로 나타나는 특징을 보인다. 장 개시 초반에는 외국인과 기관투자자의 참여가 낮은 종목이나 규모가 작은 종목에서 가격발견이 빠르게 이뤄지는 반면, 장 후반에는 외국인과 기관투자자의 참여 가 높거나 변동성이 낮은 종목, 또는 대기업 종목이 가격발견을 주도한다. 셋째, 시가 단일가매매 후반부는 초반 오버슈팅한 WPC에 조정 국면이 일어나 시가라는 균형가격 발견을 촉진하는데 이 시간대에서 외국인과 기관투자자의 호가제출 비중은 신규/정정 /취소호가에서 모두 WPC의 횡단면 결정에 통계적으로 강한 양(+)의 요인으로 작용한 다. 이는 외국인과 기관투자자가 정보거래자로서 전략적으로 행동하며 시가단일가매 매의 가격발견에 주도적인 영향을 끼치는 것을 뜻한다. 또한 단축 조치로 시가단일가 매매시간이 짧아졌음에도 외국인투자자의 가격발견 속도에 미치는 영향력은 더욱 커 졌다. 이상의 분석 결과들은 단축 이전과 이후 대체로 유사해 단축 이전 장 초반 30분은 노이즈가 많은 불확실성을 내포한 시간임을 알 수 있어 KRX 유가증권시장의 제도 개선 효과는 긍정적이었음을 시사한다.

On April 29, 2019, the Korea Exchange (KRX) reduced the duration of the pre-market opening call auction—referred to as the opening call auction session—from 60 minutes to 30 minutes in its KOSPI market. This study examines the impact of this policy change by analyzing how foreign and institutional investors contribute to price discovery during the shortened auction period. Specifically, we divide the opening call auction session into 5-minute intervals and investigate investor-specific order submission patterns and the cross-sectional determinants of Weighted Price Contribution (WPC), both before and after the change. Using data from all submitted orders over 50 trading days before and after the event, we find the following key results: First, order submission behavior differs markedly across investor types. Retail investors predominantly submit their orders during the early part of the auction session, whereas foreign and institutional investors tend to concentrate their order submissions toward the latter part of the session. Moreover, the proportion of revised and canceled order by foreign and institutional investors increases significantly near the end of the session, suggesting active price adjustment behavior. This implies that these investors behave as informed traders, delaying order submission to the final stages in order to avoid premature information disclosure and to exert greater influence over price determination of the latter part of the session. As a result of these strategic behaviors, foreign and institutional investors exhibit significantly higher order execution rates, reinforcing their dominant role in the price discovery process. Second, the determinants of Weighted Price Contribution (WPC)—used to measure the contribution of submitted orders to the opening price—vary across time segments within the call auction and reveal contrasting dynamics between the early and late phases. Specifically, the initial stage of the auction (the “WPC initial computation interval”) is the first opportunity for market participants to incorporate overnight information that has become available since the previous day’s close. During this phase, price discovery tends to occur more rapidly in stocks with lower participation by foreign and institutional investors, smaller firm sizes, higher volatility, or greater liquidity. In contrast, the latter part of the call auction (the “WPC adjustment interval”) is characterized by a reversal in these determinants. In this phase, price discovery is primarily driven by stocks that have higher participation from foreign and institutional investors, lower volatility, or larger firm sizes. This suggests that investor composition, and stock characteristics exert different influences on price formation depending on the stage of the auction. Interestingly, at the moment of opening price determination—the final order-matching point—price discovery again accelerates in stocks with lower foreign investor participation or smaller firm sizes, diverging from the characteristics observed during the immediately preceding phase. This reversal implies a process similar to a tatonnement mechanism (as in a Walrasian auction), wherein the market undergoes a trial-and-error adjustment process to reach an equilibrium price. Third, the adjustment of WPC in the latter stage of the auction session is primarily driven by the strategic behavior of foreign and institutional investors. Initially, WPC often reflects overshooting caused by overreactions from retail investors, who act as noise traders. As the auction progresses and approaches the final price determination point, WPC declines—sometimes turning negative— indicating a correction process toward a more accurate equilibrium price. During this adjustment phase, foreign and institutional investors shift from a passive stance in the early phase to an active role in the latter part of the session, submitting a large proportion of new, revised, and canceled orders that play a crucial role in correcting earlier mispricings. This behavior is empirically supported by the statistical analysis showing that the share of foreign and institutional orders during this phase exhibits a strong and positive correlation with cross-sectional WPC values, reinforcing their pivotal role in leading the price discovery process. The overall analysis demonstrates that these patterns hold consistently before and after the 2019 policy change, in which the KRX reduced the length of the opening call auction from 60 minutes to 30 minutes. This consistency implies that the first 30 minutes of the original one-hour session may have contained a high degree of market noise and uncertainty, resulting in inefficiencies in the price discovery process. The shortening measure of the opening call auction period did not diminish the role of foreign and institutional investors; rather, their strategic contributions to price discovery remained intact. These findings suggest that the policy change had a positive effect on market efficiency of this session. In conclusion, the differentiated behavior of investor types, the shifting determinants of WPC, and the strategic timing of informed traders’ order submissions together illustrate the complex dynamics underlying price discovery in a compressed opening call auction environment.

This study analyzes the stock market reaction to share sale announcements made by venture capital firms (VCs) and tests the monitoring role of VCs on publicly listed companies. We hypothesize that the divestment of shares by a VC signals the termination of its monitoring function, thereby inducing a negative market response. This effect is anticipated to be more pronounced when the divesting VC possesses stronger incentives to monitor the portfolio company. This study employs the ownership stakes and board representation of VCs as proxy variables for monitoring incentives. Analyzing 298 disclosures of share sales by VCs in the KOSDAQ market from 2010 to 2022, this study finds the following. First, a statistically significant negative cumulative abnormal return is observed from the disclosure date to five days afterward. Second, univariate analyses of subsamples indicate that the negative market reaction is more pronounced when the VC holds a larger ownership stake or a board seat. Third, regression analyses controlling for firm and VC characteristics confirm a negative correlation between the VC’s monitoring incentives and post-disclosure cumulative abnormal return, providing support for the monitoring hypothesis.

I find that a more concentrated debt structure increases the likelihood of newly issued loans being sold in the secondary market. A one-standard-deviation increase in the Herfindahl-Hirschman Index of the seven debt types prior to loan issuance is associated with a 7% higher likelihood of a new loan being sold relative to the sample average. A more concentrated debt structure enhances coordination and bargaining power among existing creditors, which discourages the participation of new lenders and weakens their bargaining position. Consequently, new entrants tend to prefer more tradable loan structures and rely on active secondary markets to mitigate these constraints. This effect is less pronounced for loans involving relationship lenders or those structured as credit lines, which can mitigate conflicts between existing and prospective creditors. By contrast, the effect is more pronounced for firms with a higher proportion of secured debt, as the presence of collateral strengthens incumbent lenders’ incentives to monitor borrowers and enforce their claims in the event of default. Moreover, the effect is amplified among financially distressed firms, where disputes over asset distribution are more likely to occur. These results underscore the critical role of a borrower’s existing debt structure in shaping investor behavior.