Research

Abstract. This paper examines the informational role of analyst reports, focusing on detecting and resolving credibility concerns that arise when recommendations diverge from the underlying content. Leveraging Large Language Models (LLMs), it develops and validates a multidimensional framework that evaluates the factual and linguistic dimensions of report content and maps them onto a unified latent scale for direct comparison with recommendations. Using more than 40,000 Chinese reports from 2018 to 2023, the analysis first confirms the informational value of both recommendations and content, and further shows how factual and linguistic content individually and jointly shape the communication of information. Taking the pre-2022 recommendation–content relationship as a benchmark, divergence is quantified and shown to elicit negative market reactions. Building on this insight, two content-based trading signals are constructed, with the pure content-derived signal consistently outperforming analyst-based signals across regimes. Overall, the study demonstrates that multidimensionally evaluated content contains bias-adjusted investment information, that credibility is dynamic and measurable, and that LLM-based analysis offers a practical way to restore transparency and unlock the information embedded in financial text.

Keywords: Analyst reports; Informational friction; Large Language Models; Chinese stock market; Asset pricing

[Download]

Abstract. Despite growing attention, the role of fiscal decentralization (FD) in managing health crises remains inconclusive. The COVID-19 pandemic has underscored the urgency of understanding the impact of FD, bringing this issue to the forefront of research. This paper investigates the effect of FD on the effectiveness of government containment measures in managing the COVID-19 health crisis across OECD countries. Results show that a higher degree of FD is associated with lower effectiveness of government policies in reducing mortality in the first phase of the pandemic. Extended analysis indicates that government containment policies are the primary channel through which FD influences mortality outcomes. Through this channel, the negative effects of FD are significantly more pronounced in countries with higher levels of governance. The first-phase findings highlight the coordination challenges inherent in decentralized systems, which contribute to the reduced effectiveness of crisis management. However, in the second phase, FD demonstrates a positive effect on the effectiveness of vaccination in improving mortality outcomes. This finding highlights the provision efficiency of public goods and services in decentralized systems. Policymakers should carefully consider the nuanced effects of FD when designing strategies for managing future health crises.

Keywords: Fiscal decentralization; Covid-19; OECD countries

[Download] 

Abstract. We investigate the cross‐section asset‐pricing patterns of major cryptocurrencies from 2017 to 2021. We show that the basis, momentum, and basis–momentum factors earn statistically significant excess returns, a result consistent with the findings reported in the commodity futures literature. The basis is the strongest signal predicting cross‐sectional differences in cryptocurrency futures returns; the momentum‐induced risk premium is not statistically powerful, whereas the basis momentum‐induced risk premium disappears when accounting for the basis‐induced risk premium. Daily factor returns are statistically much stronger than weekly factor returns. Monthly factor returns are nonsignificant. 

Keywords: Cryptocurrency futures; Asset pricing; Risk factors

[Download] 

Abstract. Hong Kong is renowned for its iconic skyline, yet the intricate details of its skyscraper development remain underexplored. Shaped by a series of historical events, the evolution of skyscrapers in the city has followed a distinctive trajectory. This research investigates key factors, inspired by existing skyscraper theories and the unique structure of Hong Kong's housing market, that influence the development of residential skyscrapers in Hong Kong from 1990 to 2023. By employing machine learning techniques, the study identifies two-year and three-year lagged annual total skyscraper heights as significant predictors of current construction activity. The results demonstrate that growth momentum is the primary driver of skyscraper development in Hong Kong.

Keywords: Skyscrapers; Hong Kong; LASSO; Machine learning

[Download] 

Introduction. This project extends recent advances in option-based risk measurement to the cryptocurrency market. Short-maturity BTC and ETH options are used to construct a Crypto Tail Risk Index (CTRI) from the pricing of deep out-of-the-money puts. The index provides a forward-looking measure of downside tail risk and is designed to test whether option-implied crash premia predict not only lower average returns, as in equities, but also the realized occurrence of crash events, which are far more frequent in crypto markets. The research aims to clarify how option markets process extreme risks in digital assets and to develop tools for forecasting, risk management, and trading in an environment characterized by high volatility and recurrent crashes.

Keywords: Cryptocurrency options; Tail risk; Volatility forecasting