Chanho Yee

• There is mounting evidence—spearheaded by Campbell and Vuolteenaho (2004) and Cohen et al. (2009)—that market cash-flow betas outperform Capital Asset Pricing Model (CAPM) betas in explaining the cross-sectional dispersion of risk premia. Existing models account for the failure of CAPM betas and the success of cash-flow betas, but they do not clarify exactly how CAPM betas become distorted relative to cash-flow betas when moving from cash-flow space to return space. We address this question by developing a parsimonious affine equity term-structure model built on a single cash-flow factor. The model jointly determines a stock’s price-dividend ratio, duration, and risk premium, as well as its market cash-flow beta and CAPM beta, all in closed form.  The model reveals that the CAPM beta fails because unpriced idiosyncratic dividend growth-rate volatility is priced in the stock’s price-dividend ratio, which, in turn, is determined by the stock's cash-flow beta. A calibration exercise, together with supporting empirical evidence, confirms the model’s implications for the relationship between the two betas.

• Technically yes, but conceptually no. While the present value identity of Campbell and Shiller (1988) and its associated return decomposition from Campbell (1991) are convenient and intuitive, I argue that the conventional definitions of these components are conceptually misleading and empirically problematic. Theoretically, the realized stock return embeds the stochastic discount factor (SDF), its covariance with cash flows (risk compensation), realized cash flows, and, if present, mispricing. Consequently, the so-called discount rate component inevitably mixes multiple sources of variation including cash-flow-related, and moreover, the discount rate component would actually be cash flow component. Empirically, the common VAR-based implementation—especially when returns are core state variables—cannot cleanly separate the components; even with augmented models including cash flow measures, discount rate news remains contaminated by cash flow and mispricing shocks. This dissertation formalizes these identification issues, derives the conditions required for clean separation, and demonstrates via simulations and structural examples how standard methods produce biased results. The contribution is a unified framework clarifying the limits of the present value identity, the assumptions it requires, and potential guidance for more reliable empirical measurement of risk premia components.

This paper reexamines the diagnostic expectations model by relaxing the restrictions to demonstrate that the current tool for detecting over- or underreaction to news can be misleading. Building on a state-space model of firm earnings with a latent AR(1) component, we allow the diagnostic distortion parameter to take positive (overreaction) or negative (underreaction) values, directly affecting the Kalman gain in investors’ updating rules. The model provides closed-form expressions that link past earnings growth, forecasts, forecast errors, forecast revision, and stock returns, yielding three key theoretical results. First, high persistence with overreaction and low persistence with underreaction can generate identical forecast error patterns, making them observationally equivalent in standard empirical tests, such as the Coibion–Gorodnichenko regression. Second, the covariances between forecasts, forecast errors, and past performance change sign systematically with the interaction of persistence and diagnostic distortion, implying that inference on reaction direction requires joint identification. Third, pricing implications from the equity term structure model follow directly: return predictability mirrors the forecast error dynamics, with similar cross-sectional return spreads arising under both overreaction and underreaction regimes. A counterexample case from the Korean stock market, which exhibits low persistence and underreaction, empirically supports our theoretical results regarding an environment that the current literature has not considered.

• The trading volume and realized volatility of major cryptocurrencies show intraday patterns similar to those of the stock market, which only appear on trading days and have sharpened in recent years. Moreover, day and night return correlations between the cryptocurrency and the stock also become positive in recent years. Using the measures of price informativeness, I find that the crypto asset price incorporates the cash flow news of the stock market. To explain this phenomenon, I develop a model in which the liquidity trading in the two markets endogenously concentrates on the same time period due to the agents' strategic behaviors. Additionally, I find that the fundamental information of the stock moves towards the cryptocurrency market.

• This paper presents a market microstructure model to explain how traders’ strategic behavior interacts with asymmetric market closures in sequential auctions. The spikes in intraday trading volume and return volatility emerge simultaneously in all traded assets due to the accumulation of liquidity trading in that period because, for discretionary liquidity traders who can choose when to trade, it is less costly when i) more assets are traded and ii) the magnitude of liquidity trading in any asset is high. The U-shaped trading volume and return volatility, with increasing return correlation, imply that the intensified trading in the earlier period is mainly due to liquidity traders, while the latter is due to informed traders. Our findings align with the observed stylized facts in assets within the same market and across different markets in the literature.

• Accelerators, which serve distinct roles from incubators, play a pivotal role in the success of startups. Although numerous studies have explored the relationship between incubators and firm performance, only a few have examined the effectiveness of accelerators, with inconsistent findings. Moreover, due to data availability constraints, studies focusing on the impact of accelerators on startups’ financial performance are lacking. We empirically examine how startups’ financial performance is influenced by the accelerator programme provided by the Korea Technology Finance Corporation. Our findings show that startups participating in the accelerator programme experienced increased asset turnover through sales growth; however, this did not translate into an improved return on assets. Our study contributes to the literature on accelerators by utilising a dataset that provides financial variables of startups and incorporates financial performance as an additional dimension of empirical evidence.