Research Papers

Brexit, Balanced Reporting and the BBC [draft available upon request]
(with Gregory S. Crawford)

R&R, American Economic Review

Abstract: We measure whether the BBC’s reporting of the Brexit debate had the unintended effect of shifting its coverage towards the Leave campaign and estimate the causal impact of this shift on referendum voters. We measure the BBC’s Leave-Remain slant and show a sharp, statistically significant, Leave-ward turn in the runup to the Brexit vote. A novel topic decomposition shows that 72% of this shift was due to a Leave-ward shift in slant in three topics closely tied to Brexit. Using both survey and aggregate data, we find exposure to BBC News in the runup to the Brexit vote caused a 4.8 p.p. increase in Leave vote intention, a magnitude large enough to have tipped the outcome.

Mergers and Market Power
(with Paolo Mengano)

Abstract: Can mergers and acquisitions (M&A) explain recent trends in rising market power? While existing literature primarily focuses on the direct effects of mergers on acquirer markups, we propose a novel approach by examining the impact of M&A through the lens of revenue transfers. We introduce a unique methodology to quantify such transfers, revealing how they significantly shape industry composition and market power. Our analysis indicates that M&A activities, often involving substantial revenue shifts, play a crucial role in the rise of aggregate markups, explaining all of the increase in concentration and accounting for 40-80% of the markup rise. These findings offer new insights into the factors driving market power trends, emphasizing the need to consider broader revenue implications of M&As. 

Do Policies Affect Preferences? Evidence from Random Variation in Abortion Jurisprudence
(with Daniel Chen, Susan Yeh)

R&R, American Economic Journal: Economic Policy

Abstract: Turning to courts to vindicate right often led to resistance and subsequent acceptance. We formalize these effects in a model where laws can generate temporary shocks. We then exploit two layers of judge randomization to estimate effects of abortion jurisprudence using all abortion appellate and district cases from 1971-2004. Four results emerge. Judges' politics, religion, and ethnicity predict abortion verdicts. Verdicts affects state regulations that restrict abortions and impact individuals. Stated and revealed preference shift against legalized abortion, albeit briefly. Backlash effects are pronounced among Republicans and on attitudes towards discretionary abortions. An original data entry experiment replicate these patterns.

Mandatory Disclosure: Theory and Evidence from Industry-Physician Relationships
(with Daniel Chen, S. Eric Reinhart, Glen Taksler) 

Journal of Legal Studies, 2019, 48(2): 409-440.

Abstract: The interaction of disclosure laws and the targeted behavior is typically unknown since data on disclosed activity rarely exist in the absence of disclosure laws. We exploit legal settlements disclosing pharmaceutical company payments across the United States. Strong-disclosure states (requiring publicly available data) had reduced payments among doctors accepting less than $100 and increased payments among doctors accepting greater than $100. Weak-disclosure states (requiring reporting to state authorities), despite imposing administrative compliance costs to industry, were indistinguishable from nondisclosure states, which suggests physicians’ disclosure aversion as a primary mechanism. Additional analyses holding fixed the cost for pharmaceutical companies of disclosing data and a differences-in-discontinuities model in distribution of payments at the disclosure threshold among strong- and weak-disclosure states support this interpretation. Significant disclosure aversion reducing conflicts of interest is consistent with the policy goals of mandatory disclosure, though the increased payments among those receiving large payments may have been unintended.

Slant in Public Service Broadcasting: Evidence from Five PSBs
(with Gregory S. Crawford)

Abstract: We measure media slant in five Public Service Broadcasters: BBC in the UK, ARD and ZDF in Germany, and SRF and RTS in Switzerland. Using speech of the main political parties in their respective parliaments, and benchmarking again newspapers with well-established political leanings, we find that the German and the Swiss broadcasters are "middle-of-the-road", while the BBC slant shifts sharply from Labour-leaning to Conservative-leaning in the runup to the Brexit referendum. Several analyses verify the robustness of the findings.

Supply and Demand for Mortgage Lending
(with Laura Aydinyan)

Abstract: We find large supplier effects in mortgage currency choice using credit register data of all mortgage loans issued in Armenia 2010-2016. Prior literature has documented bank incentives to transfer exchange rate risk to consumers. Controlling for consumer and loan characteristics, households are more likely to choose dollar-denominated mortgages from banks with higher dollar deposit inflows. The effect is larger for foreign banks, and for relationship banking and lower income customers. We use the currency devaluation of 2014 as an exogenous shocks to identify the causal effect and also, the channel of exchange risk transmission. The IV specification identifies the supply side effect, and the transmission works through consumers who have not experiences currency devaluation on prior loans.

Causal Political Speech

(with Oda Nedregård)

Abstract: A large and growing literature measuring political and media slant takes the differences in speech from opposing political parties as reflective of party ideology. However, politicians might speak differently for many other reasons, such as geography, education or prior employment. I use the Brexit referendum to address whether shifts in electorate preferences are reflected in politicians' speeches. The Brexit referendum was a surprise outcome, not only at the national level, but also at a local level: many MPs voiced support for Remain, while their constituencies voted Leave. These are equally likely to occur in Labour or Conservative constituencies. I use this shock to measure whether MPs with non-aligned voter preferences change their speech after the Brexit referendum.

How Households and Banks Really Choose Mortgage Rate Fixation Periods
(with Christoph Basten, Benjamin Guin)

Abstract: First, using unique data on un-intermediated household mortgage applications and a non-linear specification with Bartik instrumentation, we show that tighter budgets induce households to buy more interest rate risk insurance. Households with the tightest budgets buy less insurance, as extra wealth allows for more risk- taking. Second, unique data on multiple banks’ responses to each application show that banks generally prefer longer fixation periods given term premiums and origination fixed costs, but this is moderated by deposit-dependence. Third, households applying for shorter fixation periods than predicted by machine learning models obtain fewer offers, consistent with the deviation signaling unobserved additional risk.

Work in Progress

Mergers in Production Networks

News Diversity

(with Julius Schäper)

Agenda Setting: Evidence from UK Newspapers

(with Julius Schäper)

Bias and Emotion in News Reporting

(with Julius Schäper)

Common Ownership and Product Differentiation