Research profile: My research focuses on finance, macroeconomics and topics at the intersection of these two fields. Much of my work is data-driven and combines standard econometric methods with web-based data and natural language processing (NLP). I mostly use these techniques to measure otherwise difficult-to-observe economic concepts, and to gain insights into the causal linkages that explain the observable behavior of individuals and firms.
Current Working Papers
Sectoral Media Focus and Aggregate Fluctuations (with Ryan Chahrour and Kristoffer Nimark, May 2021) (R&R, American Economic Review)
We formalize the editorial role of news media in a multi-sector economy and show that it can be an independent source of aggregate business cycle fluctuations, even when sector developments are reported accurately. In the model, news media monitor the economy on behalf of firms and make state-dependent decisions about which subset of sectors to report on. Our approach tightly links agents’ beliefs to real economic developments and allows for incomplete information without exogenous noise shocks. Accurate public reporting about a sector that is unrepresentative of the economy as a whole causes firms across all sectors to over- or underinvest in productive capacity relative to a full information benchmark. We construct a novel data set of sectoral news coverage in the US and use it together with standard input-output tables to calibrate a 29-sector model. The calibrated model can replicate several features of the data, including the severe recession in 2009, even though sectoral TFP shocks are the only source of exogenous variation.[Coverage at the Richmond Fed]
The Use of Credit Ratings in Financial Markets (with Ramin Baghai and Bo Becker, August 2020) (Reject & Resubmit, Management Science)
Fixed income markets rely on delegated asset management, where fund managers’ portfolio decisions are directed and restricted by investment mandates. We use textual analysis to classify fixed income funds’ mandate contents. Credit ratings are used by a large majority of these funds to delineate which assets may be held. Despite the shortcomings of ratings revealed by the global financial crisis, their use in mandates has steadily increased over the 1999-2018 period. This continued reliance on credit ratings suggests that ratings are an integral feature of contracting in financial markets and points to a lack of practically useful alternatives. [Blog post at Chicago Booth School of Business] [Summary at European Institute of Corporate Governance] [Coverage at the Wall Street Journal] [Coverage at Bloomberg]
During the 2008-2010 financial crisis a large number of firms experienced acute liquidity shortages and adjusted real policies such as their employment and investment decisions. Searching the texts of approximately 900,000 SEC filings from the past twenty years, we document that these reactions were very similar to the ones implemented by firms that face shortages during more normal times. However, what made the crisis unique was that many of the affected firms were large, profitable and difficult to distinguish from the median public company. As a result, many widely-used financial constraint measures lost much of their informational content during the crisis and underestimated the number of affected firms. We discuss the theoretical and empirical implications of our findings and what they say about the nature of the financial crisis.
In this paper, I use high-frequency financial market estimates to identify the monetary policy shock in a non-recursive 133 variable FAVAR. All restrictions are imposed exclusively on impact, and only on financial market variables. Using the economy’s underlying factor structure as the link between its real and financial sides, I find that high-frequency responses contain valuable information about the behavior of lower-frequency macro variables. Even though the proposed identification scheme does not fall back on any of the standard (FA)VAR identifying assumptions, it confirms the classical finding that monetary policy has strong and significant delayed effects on real activity. I also obtain stock market responses that are compatible with the efficient market hypothesis and find that consumer prices react very little to monetary policy
How Do Firms Set Prices? Narrative Evidence From Corporate Filings (2020) European Economic Review
We quantify narrative evidence from archived corporate filings to construct a novel dataset on the price-setting behavior of public companies. Our approach is closely related to that of survey-based studies on price setting, but we collect data from two decades instead one specific point in time. This allows us to investigate variation over time, and to establish that qualitative information provided directly by firms is informative about the aggregate variables macroeconomists are fundamentally interested in. Our findings suggest that (i) price-setting is asymmetric and has a strong state-dependent component, (ii) real rigidities in the form of strategic pricing complementarities are very common, and (iii) an increase in the number of price changes does not necessarily reflect a decrease in price rigidity. We discuss to what extent our results can be explained by strategic reporting behavior and consider their implications in the context of widely-used price-setting models.
News Media and Delegated Information Choice (with Kristoffer Nimark, 2019) Journal of Economic Theory
No agent has the resources to monitor all events that are potentially relevant for his decisions. Therefore, many delegate their information choice to specialized news providers that monitor the world on their behalf and report only a curated selection of events. We document empirically that, while different outlets typically emphasize different topics, major events shift the general news focus and make coverage more homogeneous. We propose a theoretical framework that formalizes this type of state-dependent editorial behavior by introducing news selection functions. We prove that (i) agents can always reduce the entropy of their posterior beliefs by delegating their information choice, (ii) state-dependent reporting conveys information not only via the contents of a story, but also via the decision of what to report, and (iii) an event that is reported by all news providers is common knowledge among agents only if it is also considered maximally newsworthy by all providers. As an application, we embed delegated news selection into a simple beauty-contest model to demonstrate how it affects actions in a setting with strategic interactions.
We use data from approximately 50,000 crowdfunding projects to assess the relative funding performance of for-profit and non-profit campaigns. We find that non-profit projects are significantly more likely to reach their minimum funding goals, and that they receive more money from the average funding provider. At the same time, however, they have fewer funding providers and obtain lower total funding amounts. Our analysis shows that these results are driven by a small number of very successful for-profit projects. We argue that the findings are consistent with a simple selection mechanism in which entrepreneurs make the non-profit / for-profit decision based on expected project payoffs.
Research in Progress
Many interesting papers...