Research

ABSTRACT: 

I show that firm-level supply-chain dependence on geoeconomic center constitutes a novel systematic risk (SCF). A zero-cost portfolio sorted on SCF earns a 1.3% monthly premium, which rises to 4.2% after the 2018 U.S.-China tariff announcements. The economic mechanism is market perceptions of geoeconomic distress, which I quantify using a LLM-based textual-analysis pipeline applied to earnings calls across 41 countries. A factor based on this perceived geoeconomic distress fully explains the SCF premium. These results indicate that geopolitical tensions can convert firm-specific supply-chain risk into a systematic risk.

with Florina Silaghi, Fangfang Wang

ABSTRACT: 

We develop ESG communication measures based on a textual analysis of quarterly earnings conference calls for U.S. public firms, distinguishing between specific and generic content. We find that managers tend to increase ESG discussions in response to ESG pressure and financial underperformance,  whereas analysts  place  less  emphasis on  ESG  matters. Managers’ ESG discussions are often generic when driven by ESG reputational risk, but become more substantive under dual pressure from shareholders and other stakeholders. Additionally, managers’ specific ESG discussions reduce analyst forecast error and increase analysts’ optimism. On the contrary, generic content increases uncertainty and dampens optimism among analysts. Managers’ excessive scripted ESG comments further reduce analysts’ optimism. Finally, our results suggest that ESG rating  agencies  incorporate  managers’ scripted  ESG  talk in  their rating  process,  whereas spontaneous ESG talk increases ambiguity, leading to greater rater disagreement. 

CONFERENCE: 

Adam Smith Sustainability Conference & 2nd Annual Conference of the British Accounting Review (2024); The 41st International Conference of the Association Française de Finance (AFFI, 2025); World Finance Conference (2025); 32nd Spanish Finance Forum (2025).

with Luca Del Viva, Carlo Sala

ABSTRACT: 

Various contentions exist regarding the return predictability of option characteristics. We propose a comprehensive option-implied predictor by aggregating 17 equity option characteristics. This predictor, identified by build-up mispricing, generates an average monthly abnormal return of around 1.3%, highlighting the equity option market's role in exacerbating stock price distortions. Notably, option implied predictor not only predict return discrepancy between option and stock market, but also predict asymmetrically future return three (nine) times higher (lower) on Earnings Announcement Days (EAD). This predictor also forecasts analysts' bias cycle and amplifies their forecasting errors over time. Managers' sentiment during earnings calls positively predicts future expectations of option traders. This paper contributes to the debate on option market efficiency and return predictability by shedding light on how mispricing, rather than risk factors, drives price deviations from fundamentals.

CONFERENCE: 

Financial Management & Accounting Research Conference (2024); 31st Spanish Finance Annual Forum (2024); Financial Management Association Annual Meeting (2024); Southern Finance Annual Meeting Annual Meeting ( 2024);

ABSTRACT: 

This paper studies whether institutions rely on option-implied equity signals for rebalancing and whether this reliance reflects skill or risk. Using a Reliance on Option-Implied information (ROPI) index from holdings and equity options (2002–2020), high-ROPI institutions, especially mutual funds, earn higher future alphas yet are more vulnerable when option infrastructure fails (e.g., the 2013 NASDAQ freeze). I introduce LS-vote, a skill-weighted consensus factor that predicts returns beyond standard models and switches sign-negative in calm periods, positive in crises. Four channels explain how option reliance skill pay: information processing, risk management, regulatory constraints, and market microstructure. Interestingly, the Volcker Rule reshaped, rather than muted, option-information use, preserving market-making capacity while policing proprietary trading can dampen crowding. Regulators can track ROPI and LS-vote as early-warning metrics and increase options-market resilience to curb tail risk without sacrificing price discovery.

ABSTRACT: 

This paper investigates how the sequencing of quantitative and qualitative corporate disclosures affects investor interpretation, market efficiency, and the efficacy of disclosure regulation. When earnings announcements and earnings calls are released simultaneously, investors anchor on the numerical earnings surprise and underreact to qualitative signals, particularly managerial uncertainty tone. This anchoring leads to exaggerated immediate price reactions followed by reversals, whereas qualitative disclosures issued in isolation yield more gradual and accurate price adjustments. These effects generate a significant and tradable 2% average monthly abnormal return unexplained by standard risk factors. The findings are stronger among retail-dominated stocks and attenuated when institutional ownership is high, suggesting heterogeneous capacity to process complex disclosures. The results raise concerns about the unintended consequences of bundled disclosures and highlight the need for regulatory attention to timing structures in corporate communication. By uncovering how disclosure formats influence belief updating, this study contributes to debates on investor protection, transparency, and optimal disclosure policy.

with Florina Silaghi, Fangfang Wang

ABSTRACT: 

We explore the effect of earnings call sentiment on the CDS market by analyzing 8,447 quarterly earnings call transcripts from 334 U.S. firms from 2014 to 2022. Our findings reveal that adjusted CDS spreads decrease by 0.619 (increase by 0.421) basis points within 2 days surrounding earnings calls in response to positive (negative) sentiment. The impact of sentiment is more pronounced for speculative-rated firms. CDS spreads also increase following a substantial sentimental disparity between the scripted management speech and spontaneous manager-analyst dialogue, or between managers and analysts. Finally, the credibility of earnings call sentiment is discounted when firms exhibit signs of earnings manipulation or ESG-related impression management.

CONFERENCE: 

Southern Finance Annual Meeting Annual Meeting (2024); World Finance Conference (2024); The 41st International Conference of the Association Française de Finance (AFFI, 2025);

with Yangyang Chen, Taoran Guo, Wenbin Hu, Han Zhou

ABSTRACT: 

This paper examines how mortgage lenders adjust credit supply in response to acute environmental shocks, focusing on major toxic chemical spills in the United States from 2010 to 2020. Using a stacked difference‐in‐differences design and matched data from the Home Mortgage Disclosure Act and the National Response Center, we find that mortgage approval rates fall significantly in spill‐affected areas, with larger declines following more severe incidents. The contraction is substantially smaller for lenders exhibiting greater environmental awareness, measured by the frequency of climate‐related discussions in earnings calls. This moderating effect is most pronounced among lenders expressing a more pessimistic tone on environmental risks. Our findings underscore the role of organizational awareness in enhancing institutional resilience and have implications for credit risk management and the design of financial policies in an era of increasing environmental hazards.

with Nak-Tak Jeong, Ki-Su Kim, SB Choi, SM Yang, SH You, JH Lee, MW Suh

Drowsy behavior is more likely to occur in sleep-deprived drivers. Individuals’ drowsy behavior detection technology should be developed to prevent drowsiness related crashes. Driving information such as acceleration, steering angle and velocity, and physiological signals of drivers such as electroencephalogram (EEG), and eye tracking are adopted in present drowsy behavior detection technologies. However, it is difficult to measure physiological signal, and eye tracking requires complex experiment equipment. As a result, driving information is adopted for drowsy driving detection. In order to achieve this purpose, driving experiment is performed for obtaining driving information through driving simulator. Moreover, this paper investigates effects of using different input parameter combinations, which is consisted of lateral acceleration, longitudinal acceleration, and steering angles with different time window sizes (i.e. 4 s, 10 s, 20 s, 30 s, 60 s), on drowsy driving detection using random forest algorithm. 20 s-size datasets using parameter combination of accelerations in lateral and longitudinal directions, compared to the other combination cases of driving information such as steering angles combined with lateral and longitudinal acceleration, steering angles only, longitudinal acceleration only, and lateral acceleration only, is considered the most effective information for drivers’ drowsy behavior detection. Moreover, comparing to ANN algorithm, RF algorithm performs better on processing complex input data for drowsy behavior detection. The results, which reveal high accuracy 84.8 % on drowsy driving behavior detection, can be applied on condition of operating real vehicles.

with Nak-Tak Jeong, Sehoon Yoo, Weon-Kyong Kim, Seok-Youn Han, Ho-Yong Lee & Myung-Won Suh

Train-tram railway vehicles implement the connection between urban tramlines and the surrounding railway network. Train-tram railway vehicles, which use existing infrastructure, can help to avoid large investments in new railways or tramlines and make interchanges between city center and surrounding cities unnecessary. However, present train-tram rail vehicle cannot carry out the integration of operating by means of high speed in intercity railways with operating on small radius of curvature in inner city tramlines. This paper aims to develop a new model for solid wheelsets train-tram railway vehicles, which will not only pass the curve of 25mR radius of curvature traveling on inner city tramlines with the speed of 18 km/h, but also can travel on straight railway with 200 km/h high speed between intercity. In this paper, a new train-tram model, including five car-body and five motor bogies with ten traction motors, is addressed. Expect as a real rail vehicle testing, this study prefer virtual simulation, which is an effective way to show the rail vehicle performance, such as ride stability, ride comfort and ride safety, by means of evaluating the dynamic characteristics of rail vehicle. Moreover, Design of Experiment (DOE) method is used to optimize solid wheelsets bogie system on improving passenger comfort, safety and stability of train-tram. Parameters of components of bogie system are tuned to minimize the derailment coefficient and the ride comfort index. The results shows that the best comfort index for passenger and minimum derailment coefficient are found. The results also show that this optimized new train-tram model is reliable and practical enough to be applied on real rail vehicle design.

with Seong-Mo Yang, Nak-Tak Jeong, Kwang-Seup Kim, Su-Bin Choi, Hyun-Soo Kim, Myung-Won Suh 

Recently, due to various environmental problems such as global warming, increasing of international oil prices and exhaustion of resource, a paradigm of world automobile market is rapidly changing from vehicles using internal combustion engine to eco-friendly vehicles using electric power such as EV (Electric Vehicle), HEV (Hybrid Electric Vehicle), PHEV (Plug-in Hybrid electric Vehicle) and FCEV (Fuel Cell Electric Vehicle). There are many driving cycles for performance evaluation of conventional vehicles. However there is a lack of researches on driving cycle for EV. This study is composed of part 1 and part 2. In this paper part 1, in order to develop urban driving cycle for performance evaluation of electric vehicles, Gwacheon-city patrol route of police patrol car was selected. Actual driving test was performed using EV. The driving data such as velocity, time, GPS information etc. were recorded. GUDC-EV (Gwacheon-city Urban Driving Cycle for Electric Vehicles) including road gradient was developed through the results of analyzing recorded data. Reliability of the driving cycle development method was substantiated through comparison of electricity performance. In the second part of this study, the developed driving cycle was compared to simulation result of the existing urban driving cycle. Verification of the developed driving cycle for EV performance evaluation was described. 

with Nak-Tak Jeong, Seong-Mo Yang, Kwang-Seup Kim, Su-Bin Choi, Maosen Wang, Sehoon You, Hyunsoo Kim, Myung-Won Suh 

Recently, due to various environmental problems such as global warming, increases of international oil prices, exhaustion of resource, a paradigm of world automobile market is rapidly changing from conventional vehicles using internal combustion engine to eco-friendly vehicles using electric power such as EV, HEV, PHEV and FCEV. Generally, in order to measure fuel consumption and pollutant emissions of cars, chassis dynamometer tests are performed on various driving cycles before actual driving test. There are many driving cycles for performance evaluation of conventional vehicles. However, there is a lack of researches on driving cycle for EV. In this study, the urban driving cycle for performance evaluation of electric vehicles was developed. This study is composed of two parts. In the part 1, the urban driving cycle'GUDC-EV (Gwacheon-city Urban Driving Cycle for Electric Vehicles)'was developed by using driving data, which were obtained through actual driving experiment, and statistic analysis with chronological table. In this paper part 2, in order to verify the developed driving cycle GUDC-EV, virtual EV platforms were configured and simulations were performed with actual driving data using In addition, simulation results were compared with existing driving cycles such as FTP-72, NEDC and Japan 10-15. 

with Xiaoguang Guo, Yanjun Wei, Zhuji Jin, Dongming Guo

As a widely used engineering material, optical glass is very difficult to process because of its high hardness and brittleness. The material removal mechanism still needs further investigation. In this paper, a mesh-free simulation method—smooth particle hydrodynamics (SPH)—is introduced and an ultra-precision cutting model for K9 glass is provided. Compared with conventional finite element method, the SPH method can effectively avoid mesh distortion when dealing with large deformation problems such as cutting. Several numerical simulations were carried out to further investigate the critical value of transition. Simulation results indicate that when cutting depth is small enough, the K9 glass can be machined in ductile mode and a relatively smooth machined surface without crack can be obtained. The chip is always in irregular and discontinuous way no matter what the cutting mode is. Three cutting tools with different rake angle were used in simulation and the effect of rake angle is analyzed.