Shihan Xie 谢诗涵

LLM Survey Framework: Coverage, Consistency, Identification, with Jing Cynthia Wu and Jin Xi

Abstract: We propose a new LLM-based survey framework with broad applications for studying expectations. It introduces three innovations beyond traditional surveys: (1) retrospective coverage, enabling analysis across decades and at higher frequencies; (2) internal consistency, fixing personas across control and treatment arms and waves, and treatments over time; and (3) clean identification, freezing the knowledge set at the survey date and introducing factual treatments released afterward. We validate it in a multi-wave randomized controlled trial of inflation expectations surveys, and recover human-comparable treatment effects. Leveraging our new framework, we extend the benchmark human survey (10 waves, 2018-2023) to over 50 waves dating back to 1990, demonstrating how LLM surveys unlock designs unattainable with human surveys.

Perceived Political Bias of the Federal Reserve, with Pei Kuang and Michael Weber

Featured in: Bloomberg, Brookings, Central Banking, CFO Dive, BFI Research Briefs

Abstract: We conduct a large-scale survey experiment involving a politically representative sample of 5,205 U.S. consumers to study how perceptions of the Federal Reserve's political stance shape macroeconomic expectations and institutional trust. The public is sharply divided regarding the Fed's perceived partisan leaning: in April 2024, most Republican-leaning consumers believe the Fed favors Democrats, whereas most Democrat-leaning consumers perceive it as favoring Republicans. Consumers who view the Fed as politically aligned with their own party (1) exhibit more optimistic views of current and future economic conditions and higher levels of trust in the Fed, (2) are more willing to pay for and actively engage with Fed communications, and (3) place greater weight on Fed announcements when updating their inflation expectations. These effects are amplified by strong in-group favoritism. Additionally, when presented with a scenario of a Trump presidency, respondents overwhelmingly anticipate the Fed aligning with Republicans, resulting in increased trust among Republican-leaning consumers but significantly reduced trust among Democrats. Although overall public trust remains stable, these findings highlight potential challenges to the Fed's perceived political neutrality and independence under the new administration. To gauge the aggregate implications, we calibrate a simple model to our survey estimates and show that eliminating polarization reduces welfare losses arising from unanchored beliefs by about one-third. 

Central Bank Communication with the Polarized Public, with Pei Kuang and Michael Weber

Featured in: Financial Times, VoxEU, BFI Research Briefs, Central Banking

Abstract: This paper examines the impact of political polarization on public trust in the Fed and its influence on macroeconomic expectations. Using a large-scale survey experiment which we fielded on President Trump's 2025 inauguration day, we study how households form beliefs about the Fed regarding its political leaning, independence, and trustworthiness. Political alignment significantly shapes perceptions: individuals who view the Fed as politically aligned report higher independence of and trust in the Fed, leading to lower inflation expectations and uncertainty. Strategic communication on institutional structure and policy objectives effectively mitigates perception biases, reinforcing the Fed's credibility and enhancing its policy effectiveness. 

Fiscal Policy: Financing and Indebtedness, with Jing Cynthia Wu, Yinxi Xie, and Ji Zhang

Abstract: We conduct a large-scale information randomized controlled trial (RCT) in five Eurozone countries to examine how households change their expectations regarding macroeconomic variables in response to a fiscal policy shock. While a positive shock to government spending is generally stimulative, we find its effects vary depending on the financing method and the country's level of public debt. When tax financing, fiscal multipliers are similar across countries with different debt-to-GDP ratios. However, a novel finding is that countries with higher debt levels exhibit smaller multipliers when fiscal policy is financed by issuing debt, and the relative effectiveness of debt-versus tax-financed spending hinges on a country's debt burden. To formalize the mechanisms behind this finding, we propose a New Keynesian model featuring fiscal discipline and proportional taxes.

Place-Based Green Industrial Policies and Labor Market Expectations, with Victoria Wenxin Xie and Xu Zhang

Abstract: Despite their growing use, little is known about how place-based policies shape expectations. We provide causal evidence that such policies influence labor market beliefs and behaviors. Using geographically targeted surveys with randomized information treatments, we show that localized policy, particularly the Energy Community Bonus Credit, significantly improves labor market outlooks, raising expected earnings growth and increasing job-search effort among the unemployed. National policy alone has limited effects. These positive responses are dampened among Republican respondents and those anticipating policy reversal. Our findings highlight the importance of localized policy design, perceived credibility, and political context in shaping responses to place-based policies.

Extreme Temperature and Low-Income Household Finance: Evidence from Payday Loans, with Victoria Wenxin Xie and Xu Zhang

Featured in: the Guardian, Politico Pro, ABC7

Abstract: Extreme temperature shocks, intensified by climate change, are increasingly frequent and severe. Yet their impact on household finances remains under-explored. Using loan-level payday loan data, we show that extreme temperature shocks lead to household financial distress, reflected in increased loan demand, reduced credit access, along with rising delinquency and default rates. We investigate the mechanisms behind this distress using borrower income data and the differences between online and storefront lenders. In the absence of federal aid for such events, the unmet demand for high-cost credit and deteriorating loan performance highlight the need for targeted policy interventions.

Stock-driven Household Attention, with Hie Joo Ahn

Abstract: We investigate the effects of stockholding on households' attention to the macroeconomy. Households' attentiveness is measured by their accuracy of inflation expectations and perceptions. Relative to non-stockholders, stockholders produce more accurate inflation forecasts and backcasts, disagree less about future inflation, and adjust their outlook more responsively to news, suggesting that stock-market participation raises households' attention. Frequent changes in stock prices incentivize stockholders to closely monitor financial markets for optimal trading, given the low cost of acquiring information. Consequently, paying attention to the macroeconomy helps hedge the risks associated with holding stocks. Therefore, attention heterogeneity driven by stockholdings can be a channel through which the distributional consequences of monetary policy are created.

Macroprudential Policy and Housing Market Expectations, with Pei Kuang, Kaushik Mitra, and Li Tang

Accepted with minor revision at European Economic Review

Abstract: This paper studies how macroprudential policy changes affect consumers’ housing market expectations and housing affordability perceptions in the United Kingdom. We conduct a large-scale online survey experiment presenting hypothetical changes of three borrower-based macroprudential tools: residential loan-to-value (LTV), buy-to-let LTV, and loan-to-income ratios. We find that policy tightening lowers house price expectations, reduces homebuying intentions, and worsens affordability assessments, while loosening has the opposite effects. The residential LTV ratio is the most effective tool. To interpret these findings, we embed our survey estimates into a dynamic model linking expectations, credit, and housing demand. The model shows that immediate belief shifts significantly amplify house price and consumption responses, highlighting the importance of expectations in the transmission of macroprudential policy. 

Navigating Fiscal Fog: Household Expectations in an Uncertain Fiscal Environment, with Junjie Guo, Li Tang, and Penghui Yin

Abstract: Using a novel survey of U.S. households with randomized information treatments, we examine how fiscal policy uncertainty affects household macroeconomic expectations and consumption decisions. We find that news about a 3 percentage point increase in government spending growth crowds out household consumption by approximately 1 - 1.5 percentage points. However, this effect is attenuated by 0.27 - 0.6 percentage points when fiscal uncertainty is introduced, primarily due to weakened government spending growth expectations. These average responses mask substantial partisan heterogeneity, as respondents from the party opposing the sitting administration show stronger initial crowd-out responses but significantly smaller reactions under uncertainty. A structural model calibrated to our survey data further reveals that fiscal uncertainty reduces the government spending multiplier primarily through weakened investment responses. Our results highlight the importance of fiscal communication in managing macroeconomic outcomes in uncertain environments.

Overconfidence in Private Information Explains Biases in Professional Forecasts, with Klaus Adam and Pei Kuang. Accepted at Journal of Monetary Economics

Abstract: We observe a rich set of public information signals available to participants in the Survey of Professional Forecasters (SPF) and decompose individual forecast revisions into those due to public information and a remainder due to residual information. We find that SPF forecasters overreact to residual information at almost all forecast horizons and for almost all forecast variables. In addition, forecasts are overly anchored to prior beliefs for all variables at all forecast horizons. We show analytically that overconfidence in private information qualitatively generates both of these features. It also implies that forecast errors correlate positively with past forecast revisions at the consensus level, but negatively at the individual level, as documented previously in the literature. Estimating Bayesian updating models on SPF data, we show that overconfidence in private information also replicates the observed patterns quantitatively. All estimated models display strong and statistically significant overconfidence in private information.

Effects of Monetary Policy on Household Expectations: The Role of Homeownership, with Hie Joo Ahn and Choongryul Yang. Journal of Monetary Economics, October 2024, Vol. 147, 103599. 

Featured in: Brookings, Interview with Faculti

Abstract: We study the role of homeownership in the effectiveness of monetary policy on households' expectations based on individual-level microdata in the U.S. We find that homeowners lower their near-term inflation expectations and optimism about the labor market outlook in response to a rise in mortgage rates, while renters are less likely to do so. We further show that forward guidance shocks lead to similar differences between homeowners and renters. Our results suggest that homeowners pay attention to news on interest rates and adjust their expectations accordingly in a manner consistent with the intended effect of monetary policy. We characterize this empirical finding with a rational inattention model where mortgage payments create an incentive for homeowners to acquire information on monetary policy, unlike renters. This housing-driven endogenous attentiveness is the key mechanism behind the compelling empirical link among homeownership, attention, and the transmission of monetary policy.

An Estimated Model of Household Inflation Expectations: Information Frictions and Implications. Review of Economics and Statistics, 2025, 107 (4), 1042–1058.

Abstract: This paper proposes and estimates a dynamic model of household inflation expectations with information frictions and time-varying parameters, where households use a Bayesian learning model to form and update inflation expectations. The model decomposes households' inflation expectation formation process into a learning component, a noisy signal component, and a measurement component. Model-implied household inflation expectations provide a robust fit for the expectation-augmented Phillips curve. As a result of time-varying inflation dynamics, households' attention to inflation is endogenous to its volatility. This insight offers explanations for the anchoring of inflation expectations during the Great Moderation.

Monetary Policy Shocks and Local Housing Prices. International Journal of Central Banking, 2025, 21 (3), 327 - 373.

Abstract: This paper studies the channels by which monetary policy shocks affect local housing prices. It first documents there is a sluggish response of housing prices, suggesting that informational frictions may be potentially important. It then develops a structural model of housing prices with information frictions, and exploits variations in housing prices across metropolitan statistical areas to estimate it. The important finding is that although households are well-informed about local demand, they are ill-informed about how monetary policy affects the local housing market. A counterfactual experiment using the estimated model implies that deviations from the Taylor rule in the early 2000s contributed to about two-fifths of the subsequent housing boom.