Professor
How to cite
Wei, Gang. 2026. "Professor." AEA RCT Registry. AEARCTR-0019359. https://doi.org/10.1257/rct.19359-1.0
Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.
* Primary investigator
Abstract
This study examines whether targeted financial information affects household demand for insurance protection, pension and long-term-care preparation, and financial information acquisition. We embed a four-arm randomized information experiment in a large household financial survey in Henan Province, China.
Approximately 10,000 adult respondents will be recruited from urban communities, workplaces, county and township locations, rural survey sites, financial-service channels, and elderly or retirement-related settings. Before the experimental intervention, respondents complete a common set of questions measuring household demographics, assets and liabilities, financial fragility, financial literacy, risk and time preferences, economic expectations, insurance participation and protection gaps, health, long-term-care experience, pension preparation, and digital financial capability.
Respondents are then assigned to one of four experimental conditions: (1) a control group receiving no additional experimental information; (2) a household shock and financial-risk information treatment; (3) a retirement and longevity-risk information treatment; or (4) an insurance consumer protection and contract-information treatment. Apart from the randomized information intervention, the questionnaire, outcome questions, response options, and ordering are identical across the four groups.
The primary outcomes measure respondents’ willingness to increase household insurance protection, willingness to pay annual insurance premiums, and likelihood of participating in a voluntary pension or long-term-care protection program. Secondary outcomes measure information-search intentions, the priority assigned to insurance protection in household financial allocation, acceptance of insurance products with transparent contractual terms, and willingness to obtain a free household risk-protection and pension-finance information handbook.
The primary analysis will follow the intention-to-treat principle and estimate the causal effect of assignment to each information treatment relative to the control group.
Intervention
The experiment contains four conditions.
Control Group: TREAT = 0
Respondents receive no additional experimental information before answering the post-treatment outcome questions.
Household Shock and Financial-Risk Information Treatment: TREAT = 1
Respondents receive brief information explaining that households may face shocks such as serious illness, accidental injury, or temporary interruption of the main source of income. Such shocks may simultaneously generate additional medical or rehabilitation expenses and reduce household income. The information also explains the basic risk-sharing function of insurance, whereby households may transfer part of the financial consequences of low-probability but potentially high-loss events to an insurer subject to the terms of an insurance contract.
Retirement and Longevity-Risk Information Treatment: TREAT = 2
Respondents receive brief information explaining that longer life expectancy may increase the number of years spent in retirement. Retirement may involve reduced or discontinued labor income while households continue to incur living, medical, and potentially long-term-care expenditures. Because longevity and future medical and care needs are uncertain, required retirement resources cannot be known with certainty in advance. The information also describes retirement savings, personal pensions, commercial pension insurance, and long-term-care protection as alternative or complementary tools for long-term retirement preparation and risk management.
Insurance Consumer Protection and Contract-Information Treatment: TREAT = 3
Respondents receive brief information explaining that insurance coverage depends on contractual features such as insured benefits, exclusions, waiting periods, deductibles, premium-payment periods, surrender provisions, and claims conditions. The information also explains consumers’ rights to information, choice, fair transactions, and lawful dispute resolution, and emphasizes the role of clear disclosure in helping consumers understand insurance products.
The substantive survey questions preceding the intervention are identical across treatment groups. The post-treatment outcome questions are also identical across groups. Only the randomized information presented immediately before the post-treatment outcomes differs across experimental conditions.
Experimental design
The study is a four-arm randomized information experiment embedded in a large household financial survey conducted in Henan Province, China.
The study plans to obtain approximately 10,000 valid survey responses. Recruitment is designed to cover respondents with heterogeneous income, age, urban-rural residence, and exposure to financial services. The target sample structure consists approximately of 2,500 urban high-income/middle-income respondents, 3,500 urban wage-earning or other ordinary residents, 2,500 county/township/rural residents, and 1,500 retired or elderly respondents. These figures represent overall sample-structure targets rather than fixed quotas for individual survey sites.
All respondents complete the same pre-treatment questionnaire. The pre-treatment module collects demographic and household characteristics, household assets and liabilities, financial fragility and economic shocks, subjective and objective financial literacy, risk and time preferences, economic expectations and uncertainty, insurance participation and protection gaps, insurance claims experience and institutional trust, health status, long-term-care experience, pension preparation, and digital financial capability.
Respondents are assigned to one of four experimental groups:
Control: no additional information;
Household shock and financial-risk information;
Retirement and longevity-risk information;
Insurance consumer protection and contract information.
The planned treatment allocation ratio is approximately 1:1:1:1.
Random assignment will be determined in advance by the research team and will not be chosen by interviewers or respondents. Randomized blocks will be used so that the four experimental versions are approximately balanced within major survey sites while preserving an unpredictable distribution sequence for field investigators.
Apart from the randomized information treatment in Section 8, the four versions of the questionnaire contain the same questions, response options, question ordering, and layout.
Interviewers are prohibited from selecting respondents’ treatment groups, changing questionnaire versions, interpreting or elaborating on the experimental information, recommending specific insurance products, or inducing particular responses.
The main statistical analysis will follow the intention-to-treat principle. Respondents will therefore be analyzed according to the group to which they were originally randomized, irrespective of whether they correctly identify the treatment in the manipulation-check question.
The principal specification will compare each information-treatment group with the control group. Where applicable, models will include predetermined randomization-stratum or survey-site fixed effects and pre-treatment covariates to improve precision. Unadjusted treatment estimates will also be reported.
Baseline balance will be assessed using predetermined pre-treatment characteristics, including age, gender, education, household income, urban-rural residence, existing insurance coverage, financial literacy, financial fragility, risk preferences, health status, and prior household long-term-care experience.
The confirmatory analysis will focus on outcomes specified prior to observing the experimental results. Analyses restricted to respondents who pass the manipulation check will be treated as supplementary rather than as substitutes for the intention-to-treat estimates.
Design details
Cited by
No citation data available for this registration yet.