Time-Zone Information Asymmetry：亞太 6 國驗證的深度研究報告
讀者互動
登入會員可按讚與收藏。
持續追蹤這項研究
收藏、追蹤與明日提醒只保存必要識別資料，可在會員中心檢視或刪除。
研究資訊不構成個人投資建議；提醒僅代表研究更新，不代表交易訊號。
Time-Zone Information Asymmetry in Asia-Pacific Equity Markets: A Comprehensive Study
[提出: 用戶（台股研究）+ Gemini（US→Japan 建議）, 執行: Claude]
Abstract
We document a robust and economically significant lead-lag relationship between U.S. equity returns and next-day Asia-Pacific local market returns, driven by overnight information asymmetry. Testing 8 Asia-Pacific local indices and 3 U.S.-listed control ETFs over 2013-2024, we find:
- 6 of 8 local markets pass the Harvey (2016) threshold (t > 3.0) for a simple 5-day SPY momentum signal
- All 3 U.S.-listed Asian ETFs fail — confirming the mechanism is timezone-specific
- Multi-OOS (5 periods), bootstrap CI [0.65, 2.24], and Bonferroni correction all pass
- Europe→Asia is weaker (r = 0.29-0.36) and a two-day chain effect does not exist
1. Introduction
The time-zone structure of global equity markets creates a natural information asymmetry: when the NYSE closes at 16:00 EST, Asia-Pacific markets have been closed for 10-15 hours. The next morning, Asia-Pacific markets open with the overnight U.S. information not yet reflected in local prices.
2. Data and Methodology
| Item | Specification |
|---|---|
| Signal | 5-day average SPY daily return (lagged 1 day) |
| Strategy | If signal > 0: hold local index. If signal ≤ 0: cash. |
| OOS | 2018-01-01 to 2024-12-31 (primary) |
| Transaction cost | 0.3% per switch (~42 switches/year) |
| Statistical test | Harvey (2016) t > 3.0 + Bonferroni correction |
3. Main Results
3.1 Cross-Market Evidence
| Market | SPY Lag-1 r | 5d Mom Net Sharpe | Harvey t | Pass? |
|---|---|---|---|---|
| Hong Kong (HSI) | +0.264 | 0.39 | 4.12 | ✓ |
| Australia (ASX) | +0.311 | 0.84 | 4.04 | ✓ |
| Singapore (STI) | +0.352 | 0.37 | 4.03 | ✓ |
| Korea (KOSPI) | +0.327 | 0.59 | 3.83 | ✓ |
| Taiwan (TAIEX) | +0.396 | 1.53 | 3.75 | ✓ |
| Japan (N225) | +0.419 | 1.23 | 3.69 | ✓ |
| Indonesia (JKSE) | +0.267 | 0.10 | 2.78 | ✗ |
| India (Sensex) | +0.244 | 0.47 | 1.05 | ✗ |
Controls (U.S.-listed ETFs, same timezone as SPY):
- EWJ (Japan): t = -0.58 ✗
- EWT (Taiwan): t = -0.81 ✗
- EWY (Korea): t = -0.25 ✗
3.2 Robustness
| Test | Result |
|---|---|
| Multi-OOS (5 periods, 2013-2024) | All positive for TW/JP/KR (no reversal) |
| Combined t-stat | TW 4.55, JP 4.60, KR 4.47 |
| Bootstrap 95% CI (Taiwan) | Net Sharpe [0.65, 2.24] |
| Bonferroni (8 markets) | All 6 passing markets survive |
| Europe→Asia | Weaker (r = 0.29-0.36), no chain effect |
| VIX intraday | r = -0.36 but redundant with SPY |
3.3 Mechanism
The information flow is one-hop (one trading day):
- SPY(t) → Asia(t+1): r = 0.26-0.42 (strong)
- SPY(t) → Asia(t+2): r ≈ 0 (no chain)
- STOXX(t) → Asia(t+1): r = 0.29-0.36 (weaker, subsumed by SPY)
The zero-overlap condition is necessary: U.S.-listed Asian ETFs (EWJ/EWT/EWY) fail because they trade simultaneously with SPY, eliminating the information gap.
4. Sector and Individual Stock Analysis (Taiwan)
| Target | Harvey t | Net Sharpe | Recommendation |
|---|---|---|---|
| 0056 (High Div ETF) | 3.44 | 1.55 | ★★★ Best |
| 0050 (Taiwan 50) | 3.25 | 1.62 | ★★★ |
| Hon Hai (2317) | 3.40 | 1.13 | ★★ |
| Cathay Financial (2882) | 3.06 | 0.97 | ★ |
| TSMC (2330) | 2.42 | 1.61 | △ (NS) |
| CHT (2412) | -1.17 | negative | ✗ (defensive, anti-correlated) |
ETFs are more robust than individual stocks. Defensive stocks (CHT) exhibit negative signal response.
5. Practical Implementation
For Taiwan investors : Check SPY 5-day average return after each trading day. If positive → hold 0050/0056. If negative → cash/money market fund. Expected ~42 trades/year, net Sharpe 1.53-1.62.
Key constraints : Daily frequency required (monthly fails). Transaction cost must be < 0.3%. Only effective for locally-traded instruments.
6. Limitations
- 7-year primary OOS (2018-2024); extended OOS 2013-2017 partially tested
- Transaction costs assumed 0.3% uniformly (varies by market)
- FX risk not hedged (shown to be small for Taiwan, R14)
- Information gap may narrow as algorithmic trading penetrates Asian markets
- The strategy exploits market microstructure, not fundamental mispricing — sustainability depends on persistence of time-zone gaps
7. Conclusion
Time-zone information asymmetry creates a universal, robust, and economically significant alpha source in Asia-Pacific equity markets. The effect is strongest where U.S. technology sector exposure is highest (Japan r=0.42, Taiwan r=0.40) and where there is zero trading hour overlap with the NYSE. This finding has implications for both retail investors (simple momentum rule) and academic understanding of international information transmission.