So, Wang & Zhang (2024) — Investor Corporate Visits and Predictable Returns
Journal of Financial and Quantitative Analysis, August 2024, manuscript 24456. Authors: Eric C So, Rongfei Wang, Ran Zhang (corresponding). The paper that defines the AIV (Abnormal Investor Visits) signal from SZSE-mandated on-site visit disclosures and shows it predicts 65–114 bps / month of cross-sectional returns on SZSE-listed firms (2012-07 to 2019-12, 108,874 firm-months).
One-line summary
Decompose quarterly on-site investor visits into expected (size, turnover, momentum, ROA) and abnormal (residual) components; the abnormal component is a positive cross-sectional alpha signal that survives Fama-French 5-factor controls and concentrates in neglected, low-volume firms ahead of positive earnings surprises.
Why this paper is canonical for the wiki
It is the cleanest example in the A-share literature of a non-price, observable resource-allocation signal: institutions voluntarily fly to a firm's HQ, and the SZSE forces the firm to disclose that visit within two trading days. The "costly vetting of underpriced firms" mechanism is distinct from analyst coverage proxies (Lee & So 2017), from event-window market reactions around visits (Cheng et al. 2019), from information-risk compensation, and from transient price pressure — and it persists for 6–10 months.
Headline numbers
- EW AIV quintile-5 minus quintile-1: 1.14% / month raw (t = 5.52), 0.66% DGTW-adjusted (t = 4.92), 1.02% FF5 α (t = 7.86).
- VW AIV quintile-5 minus quintile-1: 0.65% / month raw (t = 2.92), 0.39% DGTW-adjusted (t = 2.64), 0.71% FF5 α (t = 3.84).
- Strategy returns are 3.6–4× larger on earnings-announcement days than non-announcement days.
- Combining high-AIV / low abnormal-turnover long with low-AIV / high abnormal-turnover short: ~1.49% / month DGTW α.
- During COVID-19 (Feb 2020 – Dec 2021), the SiteVisit_Intensity × postFeb2020 interaction is significantly negative for mutual-fund Carhart-4 alphas — heavy-visiting funds are disproportionately harmed.
Methodological anchors
- AIV = residual of monthly cross-section regression: \text{LNVISIT}_{im} = \alpha_m + \beta_1 \text{SIZE} + \beta_2 \text{TURN} + \beta_3 \text{MOMEN} + \beta_4 \text{ROA} + \varepsilon_{im}.
- Travel cost computed from Baidu Map API optimal-route duration between fund and firm coordinates.
- Disclosure date used (not visit date) to avoid look-ahead; ~88.9% of visits disclosed within the mandated 2-day window.