Abstract: Earnings opportunities and downside risk have opposite implications for stock-market entry. I measure earnings distributions in the Current Population Survey and relate their upper and lower partial variances to subsequent portfolio transitions in the Survey of Income and Program Participation. Holding total variance fixed, a one-standard-deviation increase in upside composition predicts 1.6 percentage points more entry into stock and mutual-fund holdings, nearly one fifth of the sample entry rate. The upper tail predicts more entry and the lower tail less, while the dispersion component is small and imprecise. The relationship persists within industry–education groups and after accounting for realized earnings changes and cyclical income exposure. A model with a recurring participation cost shows how favorable earnings realizations can generate entry, and short participation spells, at unchanged mean and variance. Variance alone therefore misses how the composition of earnings risk relates to household market access.

Abstract: Using transaction-level data on 189,530 Chinese retail investors over the 2013–2016 boom and bust, we measure experience as counts of large realized gains and losses and examine how these counts relate to the disposition effect, the tendency to sell winners and hold losers. A two-standard-deviation increase in accumulated large losses raises the propensity to sell winners rather than losers by about 23% of the baseline effect, while an equivalent increase in gains lowers it by about 21%. Under composition-matched simulations, experience-independent selling rules do not reproduce this asymmetry. To explain both the baseline effect and this asymmetry, we build a memory-based recall model in which similarity-weighted retrieval of past outcomes guides selling. A single retrieval channel reproduces all three patterns without shifting preferences.

Abstract: Poor labor-market starts can raise the demand for precautionary balances while reducing the resources available to accumulate them. I study this tension using women's own deposits and securities in the Japanese Panel Survey of Consumers, measured consistently across marital states. The sample contains 2,632 women assigned to entry cohorts from 1985 to 2005. Comparing childhood prefectures within entry cohorts, a 0.5 lower job-openings-to-applicants ratio is associated with 438,000 yen less in later personal financial assets. The difference is concentrated in deposits and remains similar with prefecture-specific cohort trends. It also survives controls for childhood-prefecture conditions in each survey year. Annual income estimates are imprecise, so the balance-sheet association does not distinguish cumulative earnings losses from household allocation or precautionary behavior. Complementary panels do not establish that weak early careers amplify later financial losses. The central finding is an association between entry conditions and later personal liquidity, a dimension of financial position that household saving alone does not reveal.

Abstract: What does asking again add to a household-finance survey? I link two to six early intertemporal-choice responses from 3,099 Japanese households to financial wealth six to eight years after a baseline balance sheet. Compared with a flexible first answer, repeated category shares reduce held-out prediction error by an additional 1.57 percentage points of control-model loss when baseline wealth is unavailable, but only 0.008 points when it is included. The conditional gain varies across evaluation samples. A separate measurement problem remains: the displayed-rate wealth gradient depends heavily on the upper response category, and flexible category profiles reject its linear restriction. Repeated answers contain information about future reported financial position, but their predictive value depends on the information already observed and how the answers are represented.

Abstract: An employer insurance mandate can introduce a new household payment for a dependent spouse while replacing contributions already paid by another worker. We study this distinction in Japan's 2016 expansion of employee social insurance using longitudinal data on married part-time women. Employer-size exposure and pension status are fixed before the 2012 announcement. From announcement to enforcement, the large-versus-small-firm gap in membership in the ¥1.06–1.29 million earnings band falls 21.1 percentage points more for dependent spouses than for workers already contributing directly. Separately reported earnings from all jobs yield a 15.4-point estimate. Documented nonenrolled employment at 20 or more weekly hours also declines differentially. The earnings pattern reverses an announcement-period increase; net changes from the earlier baseline are less precisely estimated. Together, the findings document how the timing and distribution of adjustment under an employer mandate differ with workers' pre-existing household insurance arrangements.

Abstract: Industrial employment can facilitate marriage through stable earnings while delaying household formation when jobs, housing, and residence are allocated together. We study this tension in China's Third Front using the 1982, 1990, and 2000 population censuses. In 1982, a 13.7-percentage-point increase in the registered heavy-industry share is associated with a 4.23-point larger gap in never-married rates between ages 20–24 and 30–34 inside designated provinces than elsewhere. Predetermined siting controls reduce the differential to 2.24 points. Retrospective first-marriage histories observed in 2000 yield a similar age profile, and education shares account for 15.5 percent of a comparable young–old gap. However, comparisons of pre-campaign and build-out cohorts do not yield a robust additional marriage shift inside Third Front provinces. The evidence documents a recurring association between industrial composition and young singlehood, rather than a causal effect of the construction campaign. It distinguishes a demographic pattern among residents of industrial places from an effect assigned to people exposed to a particular industrial policy.

Preliminary abstract: Does the same investor treat a loss differently when it is hers rather than an algorithm’s? We study 100,000 investors on Alipay’s Ant Fortune platform (September 2020 to September 2024) who simultaneously hold a self-directed mutual-fund portfolio and a robo-advisory account (BangNiTou). Comparing the same investor across two accounts that differ only in who chose the holdings, we find that the response to losses does not weaken under delegation—it reverses sign. On her own account she holds and adds to losers (averaging down); on the delegated account she sells losers and chases winners. A single loss-aversion parameter cannot produce this sign flip. It follows, we argue, from delegation changing what a return means: on her own account a loss is processed through realization utility, while on the delegated account it is both a signal about the algorithm and a source of dissonance she resolves by exiting. We estimate the model structurally and quantify the return investors forgo through self-directed loss aversion. Full empirical results remain in progress.

Abstract: Would a slow-selling home benefit more from better valuation information or from access to more buyers? I combine Beijing housing transactions, asking-price histories, and an estimated search-and-negotiation model. Both high and low recorded asks accompany long selling times, but pricing history changes the low-ask relationship. The premium associated with few comparable sales largely disappears after conditioning on signed asks. Greater baseline follower interest predicts fewer subsequent cuts. The model distinguishes difficulty finding buyers from waiting for an optimistic offer. Joint estimation of prices, concessions, and duration favors persistent differences in buyer access. Across six fitted models, 25% more buyer encounters shorten expected selling time by 11.00–11.58% and raise seller continuation value. Better information has less uniform benefits. An equal 20% reduction in posterior uncertainty barely speeds selling in the combined model and lowers seller value by 11.64%; an information-only model with a close predictive score shortens selling time by 10.22% and slightly raises seller value. Information can remove the optimistic offers that make waiting worthwhile. These conditional model comparisons separate faster turnover from the seller's benefit and explain why sparse trading need not imply that better information will restore liquidity.

Abstract: Can lower prices sustain housing trade when buyers cannot acquire the school rights enjoyed by existing owners? I study a Beijing reform that preserved the previous admission rules for eligible homeowners, while new buyers faced assignment across several schools. Linking enrollment notices to housing transactions, I compare winter-to-spring changes around January 2019 with the same seasons in earlier years. Sales remain roughly flat in formerly key-school neighborhoods but rise in ordinary-school neighborhoods. After adjusting for earlier seasonal patterns, the gap in relative sales growth is 32–45 percent; transaction prices are also 4–5 percent lower relative to ordinary-school neighborhoods. A matching model explains why price reductions need not restore trade when a sale changes the rights attached to a home. The model also shows that portable school rights can facilitate exchange while reducing other families' access when school places are fixed.

Abstract: Retirement distributions can accompany job changes without an interruption of employment. Using 39,481 SIPP person-years of prior workers with reported 401(k) or thrift ownership, I find that voluntary separators holding a job throughout the year have a 1.67-percentage-point higher non-rollover rate than continuers (95-percent interval [0.68,2.67]). Their median year-on-year change in nominal all-job earnings was 11 percent. In a separate resource comparison, the voluntary-separation gap is 3.76 points below one month of prior earnings in financial assets and 0.42 points above that threshold. The 3.34-point interaction survives multiple-testing adjustment and appears in both analysis windows. These observational findings place retirement cash-outs at the intersection of job mobility and limited financial resources; they do not identify the effects of separation or the use of withdrawn funds.