We provide empirical evidence on how news about financial intermediaries' net worth impacts the aggregate economy,
using a high-frequency identification strategy. We measure "financial shocks" based on the idiosyncratic stock-price changes
of large U.S. intermediaries in a narrow window around their earnings announcements. We document significant effects
of
these shocks on the stock price and borrowing costs of nonfinancial firms, as well as on macroeconomic variables.
The
effects are more pronounced for firms with low credit ratings and when the aggregate net worth of intermediaries is
low.
Verifying results of published social sciences research is essential but expensive, costing hundreds of dollars per
study. With AI tools like ChatGPT becoming widespread, we tested whether they could help scientists check if research
findings can be reproduced. We assigned 288 researchers to 103 teams working with no AI, with AI as an assistant, or AI
leading the work with minimal human input. Human teams and AI-assisted teams performed similarly on most tasks, but
humans caught more critical errors. AI working autonomously achieved a 37% reproduction rate, making it potentially
useful for automated screening when human review is cost-prohibitive. These results nonetheless show that human
expertise remains essential for reliable scientific validation.
This paper provides empirical evidence of the importance of firm attention to macroeconomic dynamics. We construct a
text-based measure of attention to macroeconomic news and document that attention is polarized across firms and
countercyclical. Differences in attention lead to asymmetric responses to monetary policy: expansionary monetary
shocks raise market values of attentive firms more than those of inattentive firms, and contractionary shocks lower
values of attentive firms by less. Attention also mitigates the effects of macroeconomic uncertainty on firm
performance. In a quantitative rational inattention model that is calibrated with this new text-based measure,
inattention drives monetary non-neutrality. As average attention varies over the business cycle, so does the
efficacy of monetary policy.
We study the relationship between media portrayals of inflation and consumer sentiment. Using tools from natural
language processing, we uncover two competing narratives in US news coverage of inflation: the first relates
inflation to financial variables, while the second relates inflation to real variables. As inflation rose in 2021,
media increasingly emphasized the real economy. Linking inflation news to social network data from Twitter, we find
that exposure to articles emphasizing the connection between inflation and the real economy significantly reduces
sentiment, particularly in periods of high inflation. Shifting media narratives may therefore have contributed to
declining consumer sentiment in 2021.
We study the transmission of conventional monetary policy in China, focusing on the interaction between monetary and
fiscal policy given the unique institutional set-up for macroeconomic policy making. Our results suggest some
progress but also continued difficulties in the transmission of monetary policy. Similar to recent studies, we find
evidence of monetary policy pass-through to interest rates. However, the impact of monetary policy measures that are
not coordinated with fiscal policy is significantly weaker than that of coordinated measures. This suggests the need
for further improvements to the interest-rate based framework.
Firm creation is at the heart of macroeconomic growth, but entrepreneurial outcomes are highly unequal. In this paper,
we study the determinants of which entrants become large firms with macroeconomic importance. Using
employer-employee-shareholder linked Canadian administrative data based on tax records, we find that workers with higher
employment earnings are more likely both to enter entrepreneurship and to found larger firms with greater growth
potential. We incorporate this link between founders' labor-market and entrepreneurial capacities into a quantitative
model of endogenous occupational choice. In the presence of ex-ante heterogeneity, small business tax deductions induce
negative selection by drawing lower-potential entrepreneurs into entry, largely offsetting their positive effects in
alleviating financial frictions and generating a quantity-quality trade-off in the design of entrepreneurial policies.
This paper studies how the transmission of monetary policy varies with monetary policy narratives. Using an AI-based
data classification algorithm guided by macroeconomic theory, we construct directed graphs of the causal mechanisms
described in FOMC transcripts, which capture the narratives used to justify interest rate decisions. Even after
purging
these narratives of predictable components from contemporaneous macroeconomic conditions, we find substantial
variation
in narratives over time. Clustering the residual graphs yields three recurring types: an inflation narrative, a
finance
narrative, and a textbook narrative. Narrative-conditioned local projections reveal that the transmission of
monetary
policy is strongly narrative dependent, no narrative cluster exhibits the canonical joint decline in inflation and
output, and the price puzzle is narrative specific. These results suggest that standard shock measures average over
heterogeneous policy episodes and that narrative measurement provides a practical way to operationalize this
heterogeneity.
This paper studies how the distribution of media's reporting of firm news affects the macroeconomy. We document three
connected facts on media's reporting of firm news: corporate news coverage is highly concentrated among the largest
firms; equity financing and investments rise after media coverage; and yet these responses are largest among small,
rarely-covered firms. We develop a heterogeneous-firm model with a media sector that matches these facts, and use it to
quantify the aggregate effects of the distribution of media coverage. In the model, asymmetric information between firms
and investors leads to financial frictions that constrain firms' investments. Media may alleviate these information
frictions, but its effects are limited by its focus on large and financially unconstrained firms. Reallocating just 10%
of news coverage eliminates half of the output loss from information frictions, which suggests a substantial aggregate
effects of the distribution of media coverage.
We study the distribution of political speech across U.S. firms, using large language models to measure political
engagement in firms' communications. Our analysis reveals five facts: (1) Political engagement is rare. (2) It is
concentrated among large firms. (3) Firms specialize in specific topics and outlets. (4) Large firms engage in a broader
set of topics and outlets. (5) The 2020 surge in political engagement was associated with increased engagement by
medium-sized firms and a shift in political topics. These findings suggest fixed costs to political engagement and the
dominance of large firms’ views in the political space.
We study the empirical importance of narratives by linking narratives in newspapers to the sentiment of social media
users. First, we model narratives as directed acyclic graphs and show how exposure to different narratives can
affect expectations in an otherwise-standard macroeconomic model. We then measure competing narratives in news media
reports on the US yield curve inversion in 2019, using techniques in natural language processing. Linking these
narratives to data from Twitter, we show that exposure to the narrative of an imminent recession is associated with
a more pessimistic sentiment, while exposure to a more neutral narrative implies no such change in sentiment. In
addition, we find that narratives are contagious: their effects spread in the social network, even to those who are
indirectly exposed.
