Publications
It's All Relative! A Method to Counter Human Bias in Crowdsourced Stance Detection of News Articles (2022; with Ehsan-Ul Haq and Pan Hui)
(Proceedings of the ACM on Human-Computer Interaction,
Vol 6, Issue CSCW2)
Abstract:
Using human
intelligence to identify news articles' political stances is common in research
and practical applications. But human judgement can be biased and prone to errors
stemming from the comprehension of tasks and political alignment. This paper
proposes a relative rating method based on news articles' stances relative to
raters' own stances to avoid comprehension inconsistency and to control for
human bias in crowdsourced stance detection of news articles. We also show how
to use the relative ratings to construct a measure for raters' stances on a
political topic and to identify raters whose ratings are of higher quality than
others. We implement our proposed methods in an online experiment that recruits
Amazon Mechanical Turk users as raters for news articles on Gun Control. Using
the data from the experiment, we find evidence that raters' own stances on Gun
Control significantly impact ratings of related news articles, both at the
individual levels and at the aggregate levels. We also present evidence that
our relative-rating-based stance measure captures more information about raters'
actual stances than their self-reported stance does.
Credibility and Explicit Inflation Targeting (2022; with Robert G. King)
This
article belongs to a
collection of essays honoring Marvin Goodfriend.
Abstract:
In his 2004
inflation targeting manifesto, Marvin Goodfriend described US monetary policy as
implicit inflation targeting and advocated explicit targeting. Summarizing the
1965-2000 US inflation experience, he highlighted the importance of evolving
Fed credibility, which accords with our recent work using a quantitative New
Keynesian model. We define credibility as policy consistency with a publicly
announced framework and develop two lessons theoretically. First, under
explicit targeting, no conflict arises between flexible inflation targeting and
maintaining/accumulating credibility. Second, implicit targeting reduces the
effectiveness of expectations management and stabilization policy, as well as
opening the door to costly inflation scare episodes.
Creating
Confusion
(2021; with Chris Edmond) Online appendix
(Journal of Economic Theory, Vol 191, 105145)
Abstract:
We develop a model in which a politician seeks
to prevent a group of citizens from making informed decisions. The politician
can manipulate information at a cost. The citizens are rational and internalize
the politician's incentives. In the unique equilibrium of the game, the
citizens' beliefs are unbiased but endogenously noisy. We interpret the social
media revolution as a shock that simultaneously (i)
improves the underlying, intrinsic precision of the citizens' information, but
also (ii) reduces the politician's costs of manipulation. We show that there is
a critical threshold such that if the costs of manipulation fall enough, the
social media revolution makes the citizens worse off despite the underlying
improvement in their information.
Decentralization
and Political Career Concerns (2017; with Jiahua Che and Kim-Sau Chung)
(Journal of Public Economics, Vol 145,
201-210)
Abstract:
Politicians¡¯ career paths often start at some
subnational governments and end at the national one. Allocation of authorities
among national and subnational governments affects (i)
how tempting the prospects of taking national offices are, and hence how strong
bureaucrats¡¯ political career concerns are, and (ii) whether the incentives
generated by these political career concerns can be put into productive use at
subnational governments. We illustrate this tradeoff in determining the optimal
degree of decentralization using China as a case study. We also compare the
equilibrium degree of decentralization in autocracy and in democracy.
Optimal
Reputation Building in the New Keynesian Model (2016; First
Author, with Robert G. King and Ernesto S. Pasten)
(Journal
of Monetary Economics, Vol 84, 233-249)
Abstract:
We study
the optimal committed monetary policy when the private sector has imperfect
information and has to infer the central banker's
ability to commit. The optimal policy is designed to influence learning and
improve the central banker's reputation of being committed. The reputation
building implies that when a committed central banker first takes office, he
should resist the temptation to stimulate output with initially high but
declining inflation; he should reverse a missed inflation target rather than
accommodate it; and he should adopt a less accommodative inflation response to
a cost-push shock than a full commitment solution suggests.
The
Power of Whispers: A Theory of Rumor, Communication and Revolution (2016; with Heng Chen and Wing Suen)
(International Economic Review, Vol 57, Issue 1, 89-116)
Abstract:
We study how rumors mobilize individuals who
take collective action. Rumors may or may not be informative, but they create
public topics on which people can exchange their views. Individuals with
diverse private information rationally evaluate the informativeness of rumors
about regime strength. A rumor against the regime can coordinate a larger mass of
attackers if individuals can discuss its veracity than if they cannot.
Communication can be so effective that a rumor can have an even greater impact
on mobilization than when the same story is fully believed by everybody.
However, an extreme rumor can backfire and discourage mobilization.
Optimal
Policy with Credibility Concerns (2013)
(Journal of Economic Theory, Vol
148, Issue 5, 2007-2032)
Abstract:
This paper considers
a reputation model of optimal taxation in which the public is unsure about the
government type. A long-lived government can be trustworthy (meaning that it
commits to its announced tax rate) or opportunistic (meaning that it retains
the ability to change its tax rate after announcing it). Unlike in most prior
studies, the committed strategy in this model is optimally chosen by the
trustworthy type. We show that this change has significant consequences for the
equilibrium dynamics. The optimal committed strategy is found to vary with the
time preferences of the two government types, the initial reputation of the
government, and the elasticity of household production. This formulation
explains differences in policy responses across governments in the face of
similar credibility problems.
Modeling and
Forecasting Stock Return Volatility Using a Random Level Shift Model (2009; with Pierre Perron)
(Journal of Empirical Finance, Vol. 17, Issue 1, 138-156)
Abstract:
We consider the estimation of a
random level shift model for which the series of interest is the sum of a short
memory process and a jump or level shift component. For the latter component,
we specify the commonly used simple mixture model such that the component is
the cumulative sum of a process which is 0 with some probability (1-¦Á) and is some
random variable with probability ¦Á. Our estimation method transforms such a
model into a linear state space form with mixture of normal innovations, so
that an extension of Kalman filter algorithm can be applied. We estimate this
random level shifts models for volatility series, proxied by the logarithm of
the absolute returns. We do this for the S&P 500, AMEX, Dow Jones and the
NASDAQ stock market return indices. Our point estimates imply few level shifts for
all series. But once these are taken into account,
there is little evidence of serial correlation in the remaining noise and,
hence, no evidence of long memory. Once the estimated shifts are introduced to
a standard GARCH model, any evidence of GARCH effects disappears. We also
produce rolling out-of-sample forecasts. In most cases, our simple random level
shift model clearly outperforms a standard GARCH(1,1)
model and, in many cases, it also provides better forecasts than a fractionally
integrated GARCH model.
Managing Expectations (working paper version, May,
2008; with Robert G. King and Ernesto S. Pasten)
(Journal of Money, Credit and
Banking, Vol 40, Issue 8, 1625-1666)
Abstract:
The idea
that monetary policy is principally about "managing expectations" has
taken hold in central banks around the world. Discussions of expectations management
by central bankers, academics and by financial market participants frequently
also include the idea that central bank credibility is imperfect. We adapt a
familiar macroeconomic model so as to discuss key
concepts in the area of expectations management. Our
work also exemplifies a model construction approach to analyzing the dynamics
of announcements, actions and credibility which we think makes feasible a wide
range of future investigations concerning the management of expectations.
Working
Papers
Evolving Reputation for
Commitment: Understanding Inflation and Inflation Expectations (July 2026, with Robert G. King)
Abstract:
We
develop a theoretical framework and solution method for dynamic policy games in
which private agents are forward looking, learn about commitment capacity, and
face strategic policymakers with and without commitment. We showcase the
framework in a New Keynesian model of inflation, where reputation, defined as
private agents' belief that the policymaker can commit to announced inflation
targets, evolves with observed inflation outcomes and feeds back into
expectations and policy effectiveness. We characterize optimal policy that
internalizes this expectation-outcome loop, derive a recursive solution, and
calibrate the model to show that strategic expectation management is important
for understanding the joint dynamics of inflation and inflation expectations.
AI and Human Capital Accumulation: Aggregate and
Distributional Implications (June 2026, with Eunseong Ma)
Abstract:
This
paper investigates how human capital responses to anticipated advances in
artificial intelligence (AI) reshape aggregate and distributional consequences
of AI. We develop an incomplete-markets model with endogenous human capital and
asset accumulation in general equilibrium, featuring three skill sectors and
uninsurable idiosyncratic risk. AI enters as an anticipated, sector-biased
shock that narrows middle-skill wage premiums and boosts returns to top
expertise. We find that human capital responses to AI (i)
drive \textit{voluntary job polarization}, shifting workers from
the middle toward both lower and higher skill sectors; (ii) magnify AI¡¯s
positive effects on aggregate output and consumption, while dampening its
impact on employment; and (iii) alter inequality: even as polarization
increases disparities in income and consumption, precautionary saving by
middle-sector households reduces the rise in wealth inequality. In an extension
(AI+), where AI raises the human-capital threshold for high-skill jobs,
additional training and saving become concentrated among high-sector
households, further increasing wealth inequality.
What does Sovereign Borrowing Signal? Theory
and Evidence from Advanced Economies (Oct 2025, with Bowen Qu)
Abstract:
We assess the relative importance of three common sources of government private information in asymmetric-information models of sovereign debt: time preference, default costs, and future economic fundamentals. We develop a parsimonious signaling model that nests all three information structures and show that only private information about future fundamentals can generate a state-dependent signaling effect of sovereign borrowing on default risk. Using quarterly data for 20 advanced economies (2000Q2¨C2020Q3), we document a new stylized fact supporting such a state-dependent signaling effect: debt growth improves sovereign ratings in good fiscal states (primarily due to increases in short-term bonds) and lowers them in poor fiscal states (mainly due to growth in long-term loans)
Learning, Rare Disasters, and Asset Prices
(Jan 2016; with Michael Siemer)
Abstract:
We incorporate joint learning about state and parameter
into a consumption-based asset pricing model with rare disasters. Agents are
uncertain whether a negative shock signals the onset of a disaster or how much
long-term damage a disaster will cause and they update their beliefs over time.
The interaction of state and parameter uncertainty increases the total amount
of uncertainty and slows learning. Once the two types of uncertainty are both
priced in asset prices, their joint effect enables our model to account for the
level and volatility of U.S. equity returns without relying on exogenous
variation in disaster risk or any realization of disaster shock in the data
sample.
Coordinating Expectations and the Informational
Role of Policy
(under revision; with Ernesto
S. Pasten)
Abstract:
Policy has
leverage on the dynamics of self-fulfilling prophecies by distorting the
informational content of aggregate history.
Discussions
Discussion
of ¡°Scarcity of Safe Assets, Inflation, and the Policy Trap¡± by Andolfatto and Williamson