Job Market Paper

"Sentiment Booms and Information" (Online Appendix)

30 Minute Presentation (PhD EVS) Slides

Non-technical Summary Blog Post

I develop a macroeconomic model of information production in financial markets during asset price booms and busts. Agents acquire information to decide which firms to fund. In the aggregate, more precise information leads to less capital misallocation. The source of booms and busts determines their effect on information production. Productivity booms increase information production and are amplified by a fall in misallocation. Sentiment booms and busts crowd out information as private information becomes less likely to change the investment decision. Information production is constrained inefficient in the competitive equilibrium for two reasons. First, each trader produces information to extract rents from others. Second, traders fail to internalize that their information production improves the capital allocation. Looking through the lens of the model, the US dot-com boom of the late 1990s appears to have been driven by productivity, and the US housing boom of the mid-2000s by sentiment.

Booms driven by sentiment increase misallocation and decrease aggregate productivity by discouraging information production.

Booms driven by productivity decrease misallocation and increase aggregate productivity by encouraging information production.

Working Paper

"Does Dispersed Sentiment drive Returns, Turnover, and Volatility for Bitcoin?", joint with Janko Heineken

We test the theoretical predictions of the differences-of-opinion literature by analyzing the extensive online discussion on Bitcoin to build a time-varying sentiment distribution, defining disagreement as dispersion in sentiment. High disagreement is associated with negative returns, high turnover growth, and high volatility, confirming the theory's predictions. However, we do not find that an increase in disagreement increases the price, which is seemingly at odds with the theoretical prediction of disagreement leading to overpricing. As the theory predicts, the disagreement effect weakens significantly after shorting instruments were introduced at the end of 2017. Our results are economically significant: at the monthly frequency, a one standard deviation increase in disagreement leads to a 9.2 percentage points lower cumulative return over the following eight months, and the adjusted R2 of regressing contemporaneous returns on average sentiment and disagreement is 0.33.

"Consumer Privacy and the Value of Consumer Data", joint with Mehmet Canayaz (Penn State) and Roxana Mihet (UNIL)

We analyze how the adoption of the California Consumer Privacy Act (CCPA), which limits consumer personal data acquisition, processing, and trade, affects voice-AI firms. To derive theoretical predictions, we use a general equilibrium model where firms produce intermediate goods using labor and data in the form of intangible capital, which can be traded subject to a cost representing regulatory and technical challenges. Firms differ in their ability to collect data internally, driven by the size of their customer base and reliance on data. When the introduction of the CCPA increases the cost of trading data, sophisticated firms with small customer bases are hit the hardest. Such firms have a low ability to collect in-house data and high reliance on data and cannot adequately substitute the previously externally purchased data. We utilize novel and hand-collected data on voice-AI firms to provide empirical support for our theoretical predictions. We empirically show that sophisticated firms with voice-AI products experience lower returns on assets than their industry peers after the introduction of the CCPA, and firms with weak customer bases experience the strongest distortionary effects.

"Overconfidence and Information Acquisition in Financial Markets"

I develop a model in which overconfidence in the form of correlation neglect incentivizes costly information acquisition in financial markets. Traders' information has two sources of noise, one idiosyncratic and the other correlated between traders. Traders are overconfident in that they overestimate the share of idiosyncratic noise in their private information, i.e., they partly neglect correlated noise. I find that an infinitesimal amount of overconfidence is sufficient to generate trade when the private signal is exogenous and free. However, substantial amounts of overconfidence are needed when traders acquire costly information. I show that the model can be integrated into macroeconomic models and can be used to study trader heterogeneity. Finally, I consider an extension in which traders have limited resources for trading. Such funding constraints dampen the effect of new information on the price. Moreover, disagreement can affect the price level differently depending on the relative scarcity or abundance of trading capital.

Work in Progress

"Is there a bubble in the SNB stock?"

"Endogenous Leverage and Fragility"

Leverage cycles can generate large swings in asset prices and economic output, with devastating effects if leverage collapses. I develop a comprehensive framework of the nexus between leverage, information production, asset prices, and volatility. I find that two equilibria arise in financial markets with dispersed information and risk-averse lenders. The first equilibrium features high leverage and high asset prices but low information production and low volatility, whereas the second has the opposite properties. This result is obtained through a complementarity between high leverage and low information production. High leverage leads to a concentration of asset ownership in the hands of optimists, discouraging information acquisition as asset prices become upward biased. The lack of new information entering the market leads to a decline in volatility, allowing traders to borrow more against the asset, which has become virtually "safe." The model has implications for security design, endogenous fragility through information production, and financial regulation on leverage and transparency.