Research questions: (1) Immediate stock market reaction on listing/announcement days for the merging banks; (2) Presence, timing and scale of spillover effects to competitor banks.
Methods combined: Event study (MM, CAPM, Fama‑French) for CAR estimation; VAR-based Granger causality and IRFs for dynamic spillovers; PSM to estimate ATT and strengthen causal claims.
Immediate market reaction — Resona (Apr 1, 2018)
All three event models show statistically significant positive abnormal returns post-event; Market Model final CAR = +6.99%, CAPM CAR = +5.64%, Fama‑French CAR = +5.66%. Mean daily AR t‑stats range 5.02–6.22 (p<0.01).
Interpretation: sustained post-event upward CAR suggests lasting positive revaluation rather than only a brief speculative spike.
Immediate market reaction — MUFG (Oct 3, 2005)
Significant pre-event run-up and very large post-event CAR: CAPM CAR ≈ +29.72% (t ≈ 7.05***), Fama‑French ≈ +27.37% (t ≈ 6.48***), Market Model more conservative ≈ +12.01%. Market viewed MUFG formation as major value‑creating shock.
Implication: size/uniqueness of MUFG made comparable matches scarce (PSM issues).
Resona: Granger causality indicates Treatment → Control significant at lag 2 (F=3.80, p=0.02) and marginal at lag 3 (p≈0.07); IRF shows a positive competitor response one period later that dissipates quickly.
MUFG: Treatment → Control shows significance at multiple lags (lag1 p=0.03; lag3 p=0.02), suggesting a more complex, persistent spillover; IRF shows volatile initial negative reaction followed by a strong positive response peaking ~period 5 before decaying.
Key takeaway: evidence of unidirectional information spillovers from merging banks to peers, with magnitude/duration dependent on event scale.
Granger causality (selected lags) — Resona
Lags
Treatment → Control (F-stat)
p-value
1
0.18
0.67
2
3.80
0.02**
3
2.31
0.07*
Granger causality (selected lags) — MUFG
Lags
Treatment → Control (F-stat)
p-value
1
4.67
0.03**
2
2.56
0.08*
3
3.26
0.02**
Embedded IRF (Resona):
Embedded IRF (MUFG):
Causal estimate — PSM / ATT (summary)
Resona (PSM): Matching successful (86% match rate: 37/43). ATT estimate ≈ 0.00186 (average daily return diff ≈ 0.186%), bootstrap SE ≈ 0.00035, t ≈ 5.28***; treatment mean daily return 0.20% vs matched control 0.02%, supporting a causal positive effect.
MUFG (PSM): Low common support (match rate ≈ 47.6%); propensity scores skewed, making ATT unreliable—PSM struggled due to MUFG’s uniqueness. Reported ATT not robust; interpret cautiously.
PSM/ATT — Resona (key rows)
Metric
Value
ATT Estimate
0.00186
Bootstrap SE
0.00035
95% CI
[0.0012, 0.0026]
Number of Matched Pairs
37
Robustness checks & limitations
Placebo event tests returned no significant CARs (no false positives). Granger results robust to lag choice (lag=2 main, lag=3 marginal). PSM covariate balance tests passed for Resona matched sample.
For investors: large transformative bank mergers can produce substantial positive abnormal returns for acquirers; monitor cross‑bank information spillovers when sizing positions.
For regulators: sizable mergers (e.g., MUFG) can trigger prolonged market reassessment—monitor systemic linkages and potential for spillovers to regional peers.
思维导图
深度阅读
元数据与概览(引言与报告概览)
报告标题:Market Reactions and Information Spillovers in Bank Mergers: A Multi-Method Analysis of the Japanese Banking Sector;作者:Haibo Wang 与 Takeshi Tsuyuguchi;发布机构/支持来源:Texas A&M International University(作者单位注明)与 Otterbein University,并有校内研究奖资助说明(致谢)。
Resona(2018)主要发现:三种模型均显示并购后 CAR 呈持续上升,表明并购带来长期正向重估而非短暂泡沫式冲击;具体数值:Market Model 最终 CAR = +6.99%,CAPM CAR = +5.64%,Fama‑French CAR = +5.66%;对应均值日异常收益和 t 统计量分别为 CAPM mean AR 0.18% (t=5.02***), Market Model mean AR 0.23% (t=6.22***),Fama‑French mean AR 0.18% (t=5.04***),均在 1% 水平显著,说明统计上高度稳健。
支持图形解读(图 2):上左图(每日 AR)显示并购日前后日常 AR 波动,事件日后 AR 多数为正且幅度上升;上右图(CAR)显示三模型 CAR 同向并在事件后上升;下左为处理组 vs 控制组原始收益比较,下右为 t‑stat 比较柱状图强调统计显著性。图像原图如下: