Tai Lo Yeung 楊 大魯

Abstract: The literature has documented a counterintuitive positive link between earnings volatility and stock holdings, posing a puzzle to researchers. I illustrate that this link is driven almost entirely by the right tail of the earnings distribution. Using the Survey of Income and Program Participation (SIPP), I construct a tail-dependent measure that interacts individuals’ labor income growth variance with its skewness. Holding income volatility constant, investors exposed to right-tail shocks (opportunity risk) increase their equity share, while those facing left-tail shocks (disaster risk) sharply reduce it. Moreover, the tail effects are asymmetric: left-tail risk hurts virtually all investors, whereas right-tail risk benefits only a subset. These results show that higher-order moments of labor-income risk shape portfolio choice, highlight the limitations of variance-based uncertainty metrics, and offer micro-level support for more robust portfolio-choice models.

Abstract: Using transaction-level data on Chinese retail investors across a boom–bust cycle, we study how salient realized outcomes shape the disposition effect. We measure experience as counts of large realized gains or losses and find a sharp asymmetry: sequences of large gains attenuate the disposition effect, whereas sequences of large losses amplify it. To account for these facts, we develop a disciplined memory-based recall model in which similarity-weighted retrieval of past outcomes guides sell decisions. The model reproduces both the baseline disposition effect and the documented asymmetries, highlighting the critical—and asymmetric—role of experience in shaping trading behavior.

Abstract: China's housing boom surged for a decade yet deflated unevenly across market segments. I ask whether valuation uncertainty explains that heterogeneity. I merge 7.2 million online listings with matched transactions for Beijing, Guangdong, Zhejiang, and Shandong (2015–2022) and build two forward-looking metrics. First, squared residuals from a community-fixed-effect hedonic provide an ex-ante measure of price uncertainty. Second, the gap between a listing's quality-adjusted price and the prior-month community benchmark yields its expected one-month return. I show that the return-uncertainty slope turns sharply more negative after each shock, with the largest swings in top-quartile (hard-to-value) listings. Buyer-attention and supply-stock regressions indicate that thinner liquidity amplifies the pricing of risk. These results demonstrate that policy signals and credit constraints can rapidly re-price uncertainty, offering regulators an empirical gauge for monitoring speculative cycles in opaque real-estate markets.

Abstract: We examine how a large-scale state development campaign affects family formation by reshaping local labor demand. Prior work suggests that expansions in blue-collar employment can increase marriage and fertility because they create a larger pool of steady, decently paid male earners. We study China’s Third Front construction program—a short-lived but massive effort to relocate and build strategic industrial capacity in the interior—as a setting in which these forces shift sharply and potentially at different horizons. Linking city-level exposure measures to micro data on marriage and births, we document two distinct effects. During the Third Front period, men in more heavily designated areas delay marriage as local manufacturing intensity rises. This pattern contrasts with prior evidence, but is consistent with a temporary deterioration in marriage-market conditions driven by rapid industrial expansion and the resulting shift in the local sex ratio. After the program wanes, the surviving industrial base is associated with higher marriage rates. Overall, the results show that state-led industrialization can generate offsetting, time-varying effects on the marriage market, operating through abrupt changes in local population composition in the short run and improved economic opportunities in the longer run.

Abstract: China’s pre-2016 health insurance system maintained separate programs for urban and rural residents, generating persistent differences in plan generosity, risk pooling, and effective access to care. This paper studies the 2016 reform that integrated urban and rural resident insurance into a unified scheme and expanded benefits, creating a large-scale change in plan design with substantial distributional stakes. I exploit staggered implementation across prefectures and estimate difference-in-differences and event-study models to identify how more generous, more portable coverage affects health care demand and household financial risk exposure. The analysis focuses on outcomes central to modern insurance design debates: utilization (outpatient visits, inpatient admissions, and higher-tier provider use), out-of-pocket spending, catastrophic medical expenditure risk, and precautionary economic behavior. The results indicate that integration primarily operates through financial protection—compressing the right tail of medical spending risk for rural households—while also increasing care-seeking, particularly for groups previously constrained by limited coverage and weak portability such as rural-to-urban migrants. These findings highlight how unifying fragmented insurance can change both the level and distribution of medical spending risk, with implications for universal coverage design and for the role of insurance in reducing inequality and supporting labor mobility.

Abstract: I study how shifting from gross-receipts taxation to value-added taxation affects organizational form at entry. Using a new province–industry–year panel of Chinese registrations (1960–2022), I construct a province-level VAT coverage intensity—the share of local stock in VAT-covered services—based on the pre-2016 pilot scope and the nationwide expansion on May 1, 2016. A two-way fixed-effects design relates the formal entry share to coverage intensity. To address endogeneity, I use a Bartik instrument that interacts pre-2011 industry shares with time-varying national coverage by industry. The first stage is strong. IV estimates imply that a 10-percentage-point increase in coverage raises the formal entry share by 0.13 percentage points, consistent with input-credit and invoice-chain mechanisms.