Research

Citation: Buchak, Greg, Jiayin Hu, and Shang-Jin Wei. 2021. “FinTech as a Financial Liberator.” NBER Working Paper. https://doi.org/10.3386/w29448.

Abstract: Interest-rate ceilings on household deposits are a common form of financial repression in developing countries that shifts surplus from savers to banks and borrowers. Using proprietary account-level data from a leading Chinese FinTech platform, we examine how the introduction of the country's first deposit-like money-market fund reshapes deposit competition. Exploiting geographical heterogeneity in adoption rates, we show that banks with more exposed depositors experience larger deposit outflows and slower deposit growth. Exposed banks respond by issuing market-rate deposit substitutes, with minimal impact on bank profitability and risk taking. Thus, FinTech can facilitate bottom-up rate liberalization of financial repression. 

Presentations: 2025 WEFIDEV–RFS–CEPR Conference, ABFER 12th Annual Conference, ABFER Capital Market Development: China and Asia Seminar Series, Bank of Italy, China Financial Research Conference (CFRC 2021), China International Conference in Finance (CICF 2021), China International Conference in Macroeconomics (CICM 2021), Columbia Macro Lunch, Luohan Academy Webinar, NBER Chinese Economy Working Group Meeting, NYU Stern CGEB China Initiative, Peking University GSM Finance Webinar, Peking University NSD Faculty Seminar, the Peak Initiative of Digital Finance of Open Research

Media coverage: Columbia Business School       

Citation: Dong, Yanfei, Jiayin Hu, Yiping Huang, Han Qiu, and Yingguang Zhang. 2024. “The Use and Disuse of Fintech Credit: When Buy-Now-Pay-Later Meets Credit Reporting.” Working Paper. https://doi.org/10.2139/ssrn.4783923.

Abstract: We show that FinTech borrowers significantly reduce their BNPL usage when this loan information is set to be shared with banks after a credit reporting policy change. Borrowers with previous default records become more disciplined in repayment behaviors, proxied by lower default rates and overdue balances than those without such records. The disuse effect is more pronounced among younger borrowers, borrowers with higher consumption levels, and borrowers who have credit cards. Reduction in BNPL usage leads to decreased online consumption. We also find supporting evidence that Big Tech platforms’ screening and monitoring technologies (imperfectly) substitute for formal enforcement institutions.

Presentation: NBER Chinese Economy Working Group Meeting 2024, China FinTech Research Conference 2024, CCER Summer Institute 2024, ABFER-JFDS Conference on AI and FinTech, the Conference on FinTech Advances in Emerging Markets 2025, HKIMR-IMF-BIS Joint Conference on Sustaining Financial Stability amid Uncertainty, Fragmentation and Rapid Innovation, and Summer Workshop on Money, Banking, Payments and Finance 2025

Citation: Hu, Jiayin. 2022. “Money Creation in Big Tech Lending.” Working Paper. https://papers.ssrn.com/abstract=4315054.

Abstract: I present a novel model where Big Tech platforms create private money outside the regulated banking system. Big Tech platforms distinguish from other online markets by providing FinTech payment services, which are more cost-efficient than cash and traditional bank payments. I show that Big Tech's capacity for private money creation depends on the FinTech payment market share, which determines users' tendency to convert FinTech money into bank money. The growing market share of FinTech payments increases the lending capacity of Big Tech platforms by reducing the de facto fiat money reserve demanded by market participants, which may dampen the effect of monetary policy tightening.

Presentations: 35th Australasian Finance and Banking Conference (AFBC), Money Study Group at Peking University  

Citation: Chen, Yanting, Yingwei Dong, Jiayin Hu, and Yiping Huang. 2022. “Does FinTech Reduce Human Biases? Evidence from Two Quasi-Experiments.” Working Paper. https://papers.ssrn.com/abstract=4312010.

Abstract: We investigate whether FinTech alleviates human biases in lending decisions under asymmetric information. Using proprietary data from a large auto equity loan company in China, we find that nonlocal borrowers obtain a smaller loan-to-value (LTV) ratio than their local counterparts, even after controlling for the collateral value and other borrower characteristics. Using two quasi-experiments where the lender adopts different financial technologies, we find that replacing human decision-making with FinTech algorithms significantly reduces both the LTV ratio and default rate differences between local and nonlocal borrowers, mitigating lending biases against nonlocal borrowers. However, the introduction of FinTech credit scores to assist human decision-making through information provision has no impact. Our results thus demonstrate the potential of algorithms in correcting human biases and promoting financial inclusion. 

Citation: Chen, Yanting, Yingwei Dong, and Jiayin Hu. 2023. “In the Shadow of Big Tech Lending.” China Economic Review 79 (June): 101913. https://doi.org/10.1016/j.chieco.2022.101913.

Abstract: We investigate the impact of Big Tech lending on non-bank traditional lenders, which have a more overlapping clientele with Big Techs than traditional banks. Our empirical methodology exploits geographical differences in Big Tech penetration ratios and adopts the instrumental variable (IV) approach using FinTech payment adoption ratios and the distance to the Big Tech's headquarter. We find that the competition from Big Tech worsens the performance of branches facing stronger Big Tech competition by reducing the number of borrowers and the amount of loans. Moreover, branches in cities highly penetrated by Big Tech lending tighten the lending standard by reducing the loan-to-value (LTV) ratios, measured as the approved loan amount per unit collateral value, while keeping the average collateral requirement unchanged. Our findings are consistent with the cream-skimming hypothesis that Big Techs possess better screening technology and reduce the quality of borrowers applying for traditional loans. Our results document novel changes in and responses of the non-bank traditional lending business in the Big Tech era.

Citation: Hu, Jiayin, Yiping Huang, and Jialin Liu. 2024. “The Changing Face of Consumer Credit: Evidence from a Big Tech Platform in China.” Pacific-Basin Finance Journal 83 (February): 102254. https://doi.org/10.1016/j.pacfin.2024.102254.

Abstract: We investigate the changing pattern of consumer credit usage following a large, adverse shock. Using a unique dataset comprising the consumption, payment, and investment activities of nearly 100,000 users of a leading Big Tech platform in China, we find that consumers who have access to both FinTech and traditional bank credit reduce their bank credit card usage while increasing their FinTech credit usage. Moreover, FinTech credit works as a complement rather than a substitute for traditional bank credit on the same Big Tech platform, as the amount of FinTech-credit-enabled payments is much lower. This impact is more pronounced among female consumers, younger consumers, and consumers who invest more money in Big Tech platform. Hence, the rise of FinTech credit is potentially driven by (1) the consumption downgrading (i.e., shifting from large amounts and service consumption to small amounts and daily necessity consumption) in the post-COVID era, when people’s income and growth expectations have been reduced and the uncertainties they face have increased, and (2) payment convenience, as people become less likely to access banks after the outbreak of COVID. Our findings provide novel evidence regarding the relationship between FinTech and traditional bank credit and the interplay between consumption and consumer credit, with implications for consumption resilience in the post-COVID-19 era.

Citation: Hu, Jiayin, Songrui Liu, Yang Yao, and Zhu Zong. 2022. “Government Deleveraging and Non-SOE Liquidity Squeeze: Evidence from Subnational Debt and Government Contractors.” Working Paper. https://doi.org/10.2139/ssrn.4296428.

Abstract: We demonstrate how government deleveraging causes corporate distress in a distorted financial market. Our difference-in-differences (DID) analysis exploits China's top-down deleveraging policy in 2017, which targets shadow bank financing and reduces local governments' borrowing capacity. We find that after the government deleveraging, private firms with local government procurement contracts experienced larger accounts receivable increases, larger cash holdings reductions, and higher external financing costs. These firms also experienced more share-pledging activities by controlling shareholders, greater likelihoods of ownership changes, and deteriorated performance. We do not find similar effects among state-owned enterprises (SOEs), which enjoy funding privileges in China's financial system. 

Presentations: NBER Conference on the Chinese Economy at CUHK-Shenzhen, China Banking and Corporate Finance Conference (CBCF), CFRN Annual Conference, Tsinghua SEM, SUSTech, Asian Meeting of the Econometric Society (AMES) in Beijing, UIBE School of Banking and Finance, China Center for Economic Research (CCER) Summer Institute

Citation:  Hu, Jiayin, Wenwei Peng, and Yang Su. 2024. “Political Influences and Court Information Production: Evidence from the Municipal Debt Market.” Working Paper. https://doi.org/10.2139/ssrn.4750742.

Abstract: This paper finds that reducing political influences over local courts tightens municipal debt capacity and constrains municipal spending. The majority of local government lawsuits are with business partners, often over government payment delays. Our analysis shows that while the direct effect on monetary penalties is minimal, reduced court favoritism: 1) exposes government credit risks as payment delays without court support signal government financial distress; and 2) reduces creditors' implicit seniority by prioritizing timely payment to business partners. These effects exist only with government payment delay lawsuits, highlighting the content-specific effects of judicial rulings on the financial markets.

Presentation: CUHK Business School Brown Bag Seminar, ABFER Capital Market Development: China and Asia Webinar, China International Conference in Finance 2025, China Financial Research Conference (CFRC 2024) at Tsinghua PBCSF, the 17th Chinese Economy Summer Institute at Peking University, the Sixth Annual Conference of Government and Economics at Tsinghua University, and the 2024 CIRF-CFRI Joint Conference

Citation:  Hu, Jiayin, Xiaokang Hu, Yuchao Peng, and Yingguang Zhang. 2024. “The Green Costs of Debt Overhang: Evidence from Local Government Debt Restructuring.” Working Paper. 

Abstract: Outstanding public debts constrain a government's ability to address climate and environmental issues. We investigate the green impact of debt overhang using China's local government debt restructuring in 2015. Cities with previously heavier debt burdens subsequently experience larger improvements in air quality after debt restructuring. We also find larger increases in environmental penalties and more green patents for polluting firms. Within a city, the impact is more pronounced in areas with higher initial pollution. Alternative factors such as environmental inspection movements and other city-level characteristics cannot fully explain our results. Our paper offers novel supporting evidence for ``debt-for-climate/nature" swap arrangements.

Presentation: China Banking and Corporate Finance Conference (CBCF 2023), CCER Summer Institute 2024, NCER-CCER Conference on Chinese Economy, Peking University, Tongji University

Citation: Hu, Jiayin, Laura Xiaolei Liu, Chloe Yue Liu, Hao Qu, and Yingguang Zhang. 2025. “CEO Turnover, Sequential Disclosure, and Stock Returns.” Review of Finance 29 (3): 887–921. https://doi.org/10.1093/rof/rfaf015.

Abstract: We document that firms experience large negative stock returns during, and positive returns following, the first informational events after forced CEO turnovers. This V-shaped return pattern is driven by the strategic sequential disclosure of bad news and good news, aligned with incoming CEOs’ incentives to manage expectations. The pattern is more pronounced when these incentives are stronger, such as when firms earn higher stock returns and have higher valuation uncertainty leading up to the informational events. Evidence from firms’ earnings surprises, analysts’ forecast revisions, and large language model-based measures of disclosure behavior indicates that incoming CEOs often initially release bad news about realized and short-term earnings, projecting a broadly pessimistic outlook for the firm’s future performance, and subsequently disclose favorable news about longer-term earnings prospects. Our findings suggest that investors make the costly mistake of failing to discern the incentives behind managers’ disclosure.

Abstract: We examine the transmission efficiency of economic policies from the central to local governments using textual analysis of hand-collected three-tier government work report (GWR) data in China. We find that local officials' political agenda is disproportionately shaped by central economic policies in the year when they take office. As a result, central policies transmit better to cities with newly appointed governors and persist longer in these cities. This policy path dependence manifests into a cohort effect: local governors who assume office in the same year focus on a similar set of policies going forward. Our paper provides novel empirical evidence on the transmission mechanism of economic policies in a political hierarchy.

Presentations: CCER Summer Institute, NSD Political Economy and Development Group, NSD Political Economy Workshop

Citation: Cheng, Zijun, Jiayin Hu, and Beichen Huang. 2024. “Bankruptcy Efficiency and Corporate Debt Maturity.” Working paper. https://papers.ssrn.com/abstract=4975403.

Abstract: Short debt maturity is commonly viewed as an enforcement device when debt financing involves multiple creditors. We show that a more efficient bankruptcy system crowds out such needs and promotes long-term financing. By exploiting the staggered rollout of bankruptcy reforms, we find larger increases in bankruptcy cases and decreases in case duration when a city establishes its first bankruptcy tribunal. Furthermore, financially distressed firms file for bankruptcy earlier and preserve more assets for distribution. This shift in insolvency resolution primarily benefits creditors with longer maturities, thereby improving firms' access to long-term financing and increasing investment and employment. 

Citation: 胡佳胤,姚洋,宗铸等.国企违约与市场纪律——来自地方国企债券违约的证据[J].经济学(季刊),2024(02):395-411.

Hu, Jiayin, Yang Yao, and Zhu Zong. “SOE Defaults and Market Discipline: Evidence from the Chinese Bond Market .” China Economic Quarterly 24 (2): 395–411. https://doi.org/10.13821/j.cnki.ceq.2024.02.04.

摘要：本文利用2014-2021年中国债券市场数据研究国企违约对市场纪律的强化作用。研究发现地方国企债券违约引起同省其他地方国企债券发行总量下降接近50%，并对同省城投债发行产生溢出效应。市场纪律的强化是地方国企债券发行下降的原因：地方国企债券违约引起同省其他地方国企信用利差上升接近50个基点，投资者感知的信用风险上升约束了地方国企债券发行。本文分析表明国企违约打破了隐性担保预期，增强了投资者对国企信用风险的敏感度，强化了国企借债的市场纪律。

Abstract: This paper investigates the impact of defaults of state-owned enterprises (SOEs) on strengthening bond market discipline. We find local SOE bond defaults resulted in a decrease of nearly 50% of the total issuance of local SOE bonds in default provinces and had a significant spillover effect on the bond issuance of LGFVs. The Strengthening of market discipline is behind the decline of local SOE bond issuance: the defaults of local SOEs have caused the credit spread of SOEs in the same province to rise by nearly 50 bp, and the increase in credit risk has restrained the bond issuance. Our findings show that defaults of local SOEs help break the market expectation of implicit government guarantees and strengthen bond market discipline.

Citation: 胡佳胤,黄北辰,向昊天,张英广等.政策导向、官员变更与企业破产——来自裁判文书的证据[J].经济学(季刊),2024(01):101-118.

Hu, Jiayin, Beichen Huang, Haotian Xiang, and Yingguang Zhang. 2024. “Policy Orientation, Political Turnover, and Corporate Bankruptcy: Evidence from Court Judgments.” China Economic Quarterly 24 (1): 101–18. https://doi.org/10.13821/j.cnki.ceq.2024.01.07.

摘要：企业破产是企业退出市场的重要方式，具有优化资源配置和推动高质量发展的积极意义。本文首次通过企业破产裁判文书的微观数据探究了政治激励和官员变更对地方推动企业破产改革的影响。我们利用2016年企业破产改革成为中央政策目标这一政治激励以及各城市市委书记变更的年份差异进行多期双重差分分析，发现企业破产数量在2016年后有新市委书记上任的城市增幅更大，当地破产管理人协会设立加速，企业破产申请被地方法院受理的概率增高，且这些结果在新任市委书记来自外地或上级单位的城市中更为显著。市委书记变更对企业破产的影响在2016年以前并不显著，反映出中央政策导向对地方推行企业破产改革的关键作用。本文揭示出政治激励与人事变动对推动我国破产制度建设的重要影响。

Abstract: Using corporate bankruptcy court filings data in China, we investigate the impact of government official turnovers on bankruptcy efficiency in a relatively weak legal institution. Corporate bankruptcy cases increased significantly after 2016, when the central government began to remove the political stigma associated with bankruptcy and provide political incentives for local officials to facilitate bankruptcy reforms. Our difference-in-differences analysis shows that cities with post-2016 political turnovers experience a larger increase in bankruptcy cases, faster establishment of bankruptcy administrators associations, and higher probabilities for bankruptcy applications to be accepted by courts. These effects are muted for political turnovers that took place before 2016 and are more pronounced in cities with non-local new officials, who tend to have fewer connections with local businesses. Our paper thus reveals the political determinants of bankruptcy institutions in a transition economy.

Citation: Hu, Jiayin, Xuan Wang, Qingxu Yang, and Junjian Yi. 2024. “Gender Disparities in the Labor Market during COVID-19 Lockdowns: Evidence from Online Job Postings and Applications in China.” Journal of Economic Behavior & Organization 223 (July):199–215. https://doi.org/10.1016/j.jebo.2024.05.011.

Abstract: We investigate the impact of COVID-19 lockdown on labor market gender differences using a unique dataset of job postings and job applicants in China. We find a “she-cession” in both the national lockdown period (the short run) and afterward (the medium run). Compared to the pre-COVID level, job postings in female (male) dominated industries on average decreased by 38.87 % (37.62 %) and 30.93 % (24.32 %) in the short and medium run, respectively, demonstrating a persistently widening gender gap in labor demand. On the labor supply side, the number of female (male) job applicants decreased by 30.28 % (27.04 %) in the short run but increased by 20.83 % (17.62 %) in the medium run, showing that females search for jobs more actively than males did when the national lockdown was lifted. Hence, the she-cession in China becomes more severe in the post-lockdown period, as more females are competing for fewer available positions. Reduced job openings in contact-intensive industries, unequal intra-household division of childcare responsibilities, and decreases in family savings all contribute to these patterns. Our results have general implications for understanding labor market gender differences in public health crises and economic downturns.

Citation: Hu, Jiayin, Hongcheng Xu, Yang Yao, and Liuyi Zheng. 2021. “Is Working from Home Here to Stay? Evidence from Job Posting Data after the COVID-19 Shock.” Working Paper. https://doi.org/10.2139/ssrn.3959407.

Abstract: We use proprietary data from a leading online job portal in China to examine the labor demand transition toward working from home (WFH) after the COVID-19 outbreak, a quasi-experiment inducing the short-run WFH take-up. We find that the increase in WFH job postings is persistent in the post-pandemic era and is more prominent among firms with less pre-COVID WFH hiring experience, consistent with the learning hypothesis. Within firms, the increase is larger in cities hit harder by COVID-19. Additionally, the WFH transition is more pronounced for postings with higher wages and stricter requirements and thus has labor market inequality implications.

Presentations: AEA Annual Meeting,  CCER Summer Institute, ICEA Working from Home conference, WEAI Annual Conference

Citation: 张丹丹, 于航, 李力行, 胡佳胤, 莫怡青, 李泓孛. 中国人工智能技术暴露度的测算及其对劳动需求的影响——基于大语言模型的新证据[J]. 管理世界, 2025, 41(7): 59-72.

ZHANG Dandan, YU Hang, LI Lixing, HU Jiayin, MO Yiqing, LI Hongbo. The Measurement of AI Exposure and Its Impact on Labor Demand in China: Evidence from Large Language Models[J].Journal of Management World, 2025, 41(7): 59-72.

摘要：大语言模型人工智能技术发展迅速。本文利用2018年1月~2024年5月之间发布的125万条在线招聘信息，基于对岗位工作任务的具体描述，构建了各职业“大语言模型人工智能技术暴露度”，估计了人工智能技术对劳动力需求以及对岗位学历和薪资等要求的影响。研究发现，在样本期内，中国劳动力市场上新增职位的大语言模型人工智能技术暴露度呈现降低的趋势；暴露度较高的职业主要是对受教育程度要求较高和薪资较高的白领职业，包括会计、编辑、销售及程序员等。基于职业层面的工具变量的回归结果显示，技术暴露度与劳动力需求呈现负相关关系，高暴露度的职业薪资增幅下降、职业内部的薪资差距变大，对教育和工作经验方面的要求也越高。本文强调，中国劳动力市场对新技术的适应性有待加强，建议在加大对大语言模型人工智能技术的研发投入的同时，降低应用门槛，鼓励企业和劳动者使用新技术。

Summary: This paper investigates the impact of rapidly developing Large Language Model Artificial Intelligence (LLM-AI) technology on the Chinese labor market. Utilizing a large dataset of online job postings, the authors construct an index of "LLM-AI exposure" for various occupations and analyze its effects on labor demand, as well as the education and salary requirements of these positions. The study highlights the need to enhance the adaptability of China's labor market to these new technologies and suggests policies to encourage their adoption while mitigating potential negative impacts.

Much of the existing literature on technology exposure relies on data and occupational classifications from the United States, such as the O*NET database. However, applying these US-centric indices directly to analyze the Chinese labor market can be problematic due to potential inconsistencies in how occupations are defined and the specific tasks they entail across countries. To address these limitations and better understand the impact of new AI technologies on China's labor market, this paper adopts a novel approach by directly utilizing the detailed descriptions of Chinese online job posts. This allows us to construct an LLM-AI technology exposure index tailored to the Chinese context. This research provides timely empirical evidence on whether and to what extent AI, particularly LLMs, is replacing human jobs in China. The study period covers the rapid advancements and increased prominence of LLMs.

The primary data source is a randomly sampled dataset of 1.25 million online job postings from Zhaopin.com between January 2018 and May 2024. The authors deploy GPT-4 to assess the exposure of each detailed work activity and task to LLM-AI technology. They then correlate the DWAs and Tasks present in each job advertisement, weighted by their importance within the job, with these established exposure scores. By aggregating these scores at the occupational level, specifically using the O*NET 6-digit occupation codes, they construct the "DWA-based LLM-AI exposure index" and the "Task-based LLM-AI exposure index."

The study yields several key findings: (1) The average LLM-AI exposure of newly posted jobs showed a decreasing trend over the sample period. (2) Occupations with the highest LLM-AI exposure are predominantly knowledge-intensive white-collar jobs such as accountants, editors, salespeople, and programmers. (3) There is a negative relationship between LLM-AI technology exposure and labor demand. (4) High LLM-AI exposure was associated with lower salary increases and a larger within-occupation salary gap. Occupations with higher LLM-AI exposure tended to have higher requirements for education and work experience. (5) City-level analysis showed that cities with higher average LLM-AI exposure experienced more significant contractions in overall labor demand, greater wage inequality, and increased demands for education and experience. The findings contrast with earlier research on AI adoption in the US, which found an increase in demand for AI-related jobs without significant negative impacts on overall employment or wages.

The major innovations of this paper include the direct construction of LLM-AI exposure using Chinese online job post data, a timely analysis of the labor market impact of LLMs, an analysis of the adaptability of the Chinese labor market, and granular analysis at both occupational and city levels. The study's findings contrast with those in the US, suggesting that LLM-AI technologies may be having a more pronounced negative impact on labor demand for highly exposed occupations in China compared to the US. This highlights the importance of understanding the specific context of technological adoption in different countries. The authors emphasize the need to strengthen the adaptability of China's labor market and implement supportive policies to navigate the changes brought about by advancing LLM-AI technologies.

Citation: Hu, Jiayin, Maggie Hu, Shangchen Li, Yingguang (Conson) Zhang, and Zheng Zhang. 2022. “Horizon Risk in Renting: Evidence from a PropTech Rental Platform.” Working Paper, July. https://doi.org/10.2139/ssrn.4060219.

Abstract: Rental contracts are often short-term, creating uncertainty for renters regarding the duration of their occupancy. We analyze the factors influencing the horizon of rental housing supply. Utilizing contract-level data from a PropTech rental platform, we find that housing market conditions significantly impact the rental supply horizon. Landlords in neighborhoods with higher housing price and rent growth offer shorter contracts and are less likely to renew expiring ones. This effect is more pronounced among younger landlords and those with multiple or highly marketable properties. We further establish causality using an exogenous policy change that differentially affects the prices of different housing units.

Presentations: 12th European Meeting of the Urban Economics Association (2023), 2023 AsRES-GCREC Joint International Real Estate Conference, 2023 Asia Meeting of the Econometric Society (AMES 2023), AREUEA International Conference Cambridge 2023, SMU-Jinan Conference on Urban and Regional Economics, China International Conference in Finance (CICF 2022), 1st Summer Meeting in Urban Economics in China, NCER-CCER Conference on Chinese Economy, PKU Guanghua School of Management Brown Bag Seminar, and Tsinghua SEM Alumni Forum

Citation: 李尚宸, 张英广, 胡佳胤, 张峥. 房价外推预期和长租需求——基于住房租赁合同数据的实证分析[J]. 金融研究, 2024, 523(1): 76-95.

LI Shangchen, ZHANG Yingguang, HU Jiayin, ZHANG Zheng. Extrapolative Expectations of Housing Price and Long-Term Rental Demand: Evidence from Rental Housing Contract Data. Journal of Financial Research, 2024, 523(1): 76-95.  http://www.jryj.org.cn/CN/Y2024/V523/I1/76

摘要：本文基于房地产投资属性视角，探究房价预期对租房需求的影响。利用北京市某大型长租平台企业2015-2019年间40万份租赁合同数据和周边二手房交易数据，我们发现，2017年北京市限购政策实施后，在房价增速更快的区域，租客到期续约的可能性更高，也即高房价增速会提升人们的长租需求；在限购政策出台前，房价增长反而会降低租客到期续约率。此外，在限购后，过去多期的房价增速以及房价增长的不确定,均与租客续约行为正相关。在长租平台续租但搬家的租客,更可能换到租金更高、品质更好的房源，呈现“以租代购”的趋势。本文结论与外推预期和诊断性预期的理论预测相一致，反映出人们对未来房价增长的预期在调控政策后趋于稳定，购房决策更为谨慎，从而提升了长租需求。本文揭示了居民房价预期对居住方式选择的潜在影响，以及买房市场对租房市场的外溢作用，对房地产相关学术研究和政策制定具有启示意义。

Summary: The demand for long-term rental housing continues to grow as an increasing number of households choose to rent instead of buying their first homes. In major cities such as Beijing, Shanghai, Guangzhou, Shenzhen, and Hangzhou, the proportion of renters among the total population reached 40% by 2022. This paper systematically examines the relationship between housing price growth expectations and long-term rental demand using a proprietary dataset comprising over 400,000 rental contracts from a long-term rental platform and approximately 500,000 second-hand housing transactions in Beijing between 2015 and 2019.

We address the empirical challenge of measuring long-term rental demand by analyzing leases through the innovative business model of a long-term rental platform. Acting as a “second landlord” in the rental market, the platform signs long-term leasing contracts with property owners, acquires properties with fixed rents, and sublets them to tenants after standardizing renovation and furnishing. To minimize vacancies and maximize profits, the rental platform prioritizes tenant renewals. Therefore, tenant renewals with the platform serve as a reliable proxy for long-term rental demand unaffected by landlords. 

To investigate the impact of housing price expectations, we exploit the implementation of the differentiated housing purchase restriction (HPR) policy in Beijing in 2017 as an exogenous shock. We find a significant negative (positive) correlation between house price growth in the tenant's neighborhood and the tenant's renewal rate before (after) March 2017. Prior to the HPR policy, during a period of rapid house price growth, a 1% increase in house price growth is, on average, associated with a 2.2% decrease in tenant renewal rate after controlling for tenant, leasing, and rental property characteristics and including neighborhood and year-month fixed effects. In contrast, following the HPR policy, a 1% increase in house price growth is significantly associated with a 1.9% increase in tenant renewal rate. Our pre-trend analysis confirms that the HPR policy is critical to reversing the impact of house price growth on long-term rental demands. 

Our results suggest that as the HPR policy stabilizes house price expectations, tenants residing in neighborhoods with faster house price growth become more inclined to choose long-term rentals as their housing solution. Tenants who continue renting from the platform are more likely to switch to new leases with higher rent and to entire rentals, indicating a trend of moving up the housing ladder as they decide to rent instead of buying. We further find that in the post-HPR period, house price growth in multiple past periods is positively related to tenant renewal rates, with stronger effects observed as the renewal time approaches. Moreover, both the volatility and dispersion of house price growth are significantly and positively associated with tenant renewal rates. 

Our empirical findings align with the predictions of extrapolative expectations theory and diagnostic expectations theory in the economics and finance literature (Malmendier and Nagel, 2016; Bordalo et al., 2018; Bordalo et al., 2020). During house price booms, tenants extrapolate recent house price information and conclude that prices will continue to rise in the future. However, the introduction of the HPR policy serves as a diagnostic signal that prompts tenants to adjust their expectation formation mechanism and recognize the potential downsides. 

Apart from housing price expectations, the HPR policy may impact tenants' demand for long-term rentals by increasing the cost of home purchases. However, the differentiated HPR policy implemented in Beijing aims to deter speculative home purchases while safeguarding reasonable home purchase demand, thus having a relatively minor impact on first-home buyers. Notably, 94% of the tenants in our sample are non-local residents with an average age of 29 years, suggesting that most of them, who are eligible to purchase their first home, are not directly affected by the tightened credit constraints of the HPR. This is supported by a heterogeneity analysis showing that the effects are not related to the age of tenants but are more pronounced among tenants from less developed provinces. These pieces of evidence collectively indicate that direct credit restrictions are not the primary mechanism driving our results.

This paper makes several contributions to the literature. First, we address the empirical challenge of measuring tenants' willingness to rent for an extended period by utilizing a unique dataset from a long-term rental platform, which ensures that the renewal information reflects tenants' decision-making rather than supply-side factors. Second, we verify the spillover effects of the housing transaction market on tenants' demand for long-term rentals, thereby enriching the research on the relationship between the housing sales market and the rental market. Third, we provide empirical evidence of the application of extrapolative expectations theory and diagnostic expectations theory in real decision-making processes regarding tenants' rental renewal choices. Our empirical findings demonstrate the crucial role of future house price growth expectations in influencing residents' demand for long-term rentals.