Embedded Files
Publications
Empirical Tests of Asset Pricing Models with Individual Assets: Resolving the Errors-in-Variables Bias in Risk Premium Estimation (SSRN)
with N. Jegadeesh, K. Pukthuanthong, R. Roll, and J. Wang
Journal of Financial Economics, 2019
To attenuate an inherent errors-in-variables bias, portfolios are widely employed to test asset pricing models; but portfolios might diversify and mask relevant risk- or return-related features of individual assets. We propose an instrumental variables approach that allows the use of individual assets yet delivers consistent estimates of ex-post risk premiums. This estimator yields unbiased estimates and well-specified tests in small samples. The market risk premium under the CAPM and the liquidity-adjusted CAPM, premiums on risk factors under the Fama-French three- and five-factors models and the Hou, Xue, and Zhang (2015) four-factor model are all insignificant after controlling for asset characteristics.
Note: This paper subsumes parts of two working papers: Jegadeesh and Noh (2015) and Pukthuanthong, Roll, and Wang (2018).
with Y. Amihud
Review of Financial Studies, 2021
Lou and Shu decompose Amihud’s illiquidity measure (ILLIQ) proposing that its component, the average of inverse dollar trading volume (IDVOL), is sufficient to explain the pricing of illiquidity. Their decomposition misses a component of ILLIQ that is related to illiquidity. We find that this component affects stock returns significantly, both in the cross-section and in time-series. We show that the ILLIQ premium is significantly positive after controlling for mispricing, sentiment, and seasonality. In addition, the aggregate market ILLIQ outperforms market IDVOL in estimating the effect of market illiquidity shocks on realized stock returns.
with Y. Amihud
Journal of Financial Markets, 2021, - Lead Article
We test the pricing of the conditional systematic risk (β) of a traded illiquidity factor IML, the return premium on illiquid-minus-liquid stocks, when its risk premium is allowed to vary over time. We find a positive and significant risk premium on conditional βIML that rises in times of financial distress, measured by the corporate bond yield spread or broker–dealer loans (including margin loans). Notably, the conditional βIML is unique in being significantly priced across individual stocks. None of the unconditional and conditional βs of Fama and French and Carhart factors is consistently and significantly priced nor are the βs of popular alternative liquidity-based factors.
Executives' Blaming External Factors and Market Reactions: Evidence from Earnings Conference Calls (SSRN)
with D. Zhou
Journal of Banking and Finance, 2022
We investigate how market participants react when corporate executives strategically blame the economy or industries for poor firm performance. In the quarters subsequent to earnings conference calls, we find that the "blame sentences", which capture executives' blaming tactics, predict negative and non-reverting abnormal returns, negative earnings surprises, and analyst recommendation downgrades. These blaming tactics also reduce the sensitivity of executives' turnover to their past performance. Our findings imply that executives strategically inject the negative information about future cash flows into the blame sentences. Market participants need to combine distinct categories of information (firm-specific vs. economy-wide) to understand the implications of blame sentences, which consumes more cognitive resources and delays their reactions to the blame sentences.
with J. A. Cookson and S. K. Moon
Review of Corporate Finance Studies, forthcoming, 2024
Speculative language in corporate disclosures can convey valuable information on firms’ fundamentals. We evaluate this idea by developing a measure for speculative statements based on sentences marked with the “weasel tag” on Wikipedia. In the 16-week test period after filing, greater use of speculative statements in 10-Ks predicts higher and non-reverting abnormal returns, more insider and informed buying, and higher news sentiment. These findings imply that managers' usage of speculative language in 10-Ks reflects voluntary disclosure of their private information about the positive prospects of events when market implications of the events are uncertain and thus have room for (re)interpretation.
Active Working Papers
with T. Chordia and B. Miao
This paper provides evidence that short-term speculation destabilizes stock prices upon the release of non-earnings news. Specifically, low latency traders (LLTs) behave as speculators and exploit naive news-chasing investors causing overpriced stocks to become even more overpriced upon the release of high sentiment news. This exacerbation of overpricing, which is subsequently reversed, creates a wedge between prices and the consensus fundamental value and hurts market efficiency. The asymmetric impact of high and low sentiment news on overpriced stocks is driven by more positive news being produced in recent years and investors paying more attention to overpriced stocks.
with E. Hong and B. Kottimukkalur
We explore how the uncertain text in 10-K/Qs, a form of soft information, affects the stock price reactions to subsequent earnings releases that contain hard and quantitative information. We find that more uncertain language in 10-K/Qs leads to stronger immediate price reactions to earnings surprises but weaker post-earnings announcement drifts. Firms with higher text uncertainty in 10-K/Qs also attract greater institutional attention and more intense trading activity by sophisticated investors around earnings announcements. These findings suggest that firms using more uncertain language in regulatory filings have higher fundamental uncertainty and attract more attention from attention-constrained investors.
We investigate whether an industry’s position in the network of inter-industry trade affects the speed of information flow. We find that return predictability to central industries from their related (=customer and supplier) industries is substantially stronger than that to peripheral industries from their related industries. Long-short portfolios of central industries yield risk-adjusted returns of 7.0% to 7.9% per annum, which are 3.6% to 5.3% higher than those of peripheral industries. To explain this finding, we argue that investors who invest in central industries need to process more complicated information about related industries, making the prices of central industries slower to incorporate all the information. We find that sell-side analysts of central industries also face more complicated information about related industries, as their earnings forecast revisions of related industries predict their future revisions of central industries more strongly. In addition, we present evidence that our finding is not explained by existing anomalies.
Page updated
Google Sites
Report abuse