台灣在2003年發行第一檔指數股票型基金 (Exchange-Traded Fund, ETF),至2023年底ETF投資人已超過700萬人。隨著龐大且穩定的散戶資金投入ETF,以追蹤特定主題或創新科技為投資策略的主題型ETF亦逐漸興起。本研究針對主題型ETF進行分析,探討其上市時機和投資績效。 本研究分析自2016年起成立的國內發行台幣計價的主題型ETF共49檔,投資國內23檔的對照組以元大台灣卓越50 ETF代表大盤,跨國投資26檔以元大S&P 500 ETF代表大盤。第一部分追蹤主題型ETF的標的指數報酬率以分析上市前後績效的表現和時機的選擇。結果顯示主題型ETF於成立前6個月和3個月,標的指數報酬率明顯高於對照組 (6.68%和4.24%),特別是跨國投資主題型或科技主題型ETF。但成立後的報酬率與對照組比較並沒有統計差異。主題型ETF的推出時機通常已過了該主題景氣循環的高峰。 第二部分探討主題型ETF上市後的投資績效,結果顯示主題型ETF與對照組的報酬率互有領先,兩組之間沒有統計上差異。兩年以上投資報酬率無法勝過對照組,而且投資時間越久,報酬率越低。進一步比較績效指標,結果顯示主題型ETF與對照組的績效指標亦無統計上差異。但是主題型ETF 的Jenson’s α與Treynor ratio高出對照組,而Sharpe ratio與β係數則低於對照組。 由於本研究的主題型ETF多數是跨國投資的科技類型ETF,因此再進行第三部分敏感性分析以測試成立前後的標的指數報酬率是否真正存在差異。方法是改變26檔跨國投資比較對象為NASDAQ 100指數,而23檔投資國內比較對象沒有變動。結果顯示49檔主題型ETF於成立前6個月和3個月,標的指數報酬率依然明顯勝出對照組 (5.44%和4.42%),特別是跨國投資主題型或科技主題型ETF。 主題型ETF成立後該主題的景氣循環高峰通常已過,因此無法再創造高於大盤的報酬與績效。除少數主題型ETF外,大部分的報酬與績效指標也無法表現持續優於大盤,而且投資越久報酬越低,不適合作為核心資產配置。
In 2003, Taiwan launched its first Exchange-Traded Fund (ETF), and by the end of 2023, the number of ETF investors had exceeded 7 million. With the substantial and steady retail investment flowing into ETFs, thematic ETFs focusing on specific themes or innovative technologies have gradually gained popularity. This study focuses on thematic ETFs, examining their listing timing and investment performance. The research analyzes 49 domestically issued thematic ETFs denominated in New Taiwan Dollars established since 2016. The 23 domestic investments benchmark for comparison is represented by the Yuanta Taiwan Top 50 ETF, tracking the overall market. For 26 international investments, the Yuanta S&P 500 ETF represents the benchmark. The first part of the study examines the benchmark index returns of thematic ETFs to analyze the performance and timing selection before and after listing. The results show that in the past 6 and 3 months before listing, the benchmark index returns of thematic ETFs are significantly higher than the comparison group (6.68% and 4.24%), especially for cross-border investments ETFs or technology-focused ETFs. However, there is no statistical difference in returns compared to the benchmark after establishment. The timing of launching thematic ETFs often occurs after the peak of the thematic economic cycle. The second part explores the post-listing investment performance of thematic ETFs. The results indicate that the returns of thematic ETFs and the benchmark alternately lead each other, with no statistical difference between the two. The investment return of thematic ETFs over a period of more than two years cannot surpass the benchmark, and the longer the investment period, the lower the return. Further comparison of performance indicators reveals no statistical difference between thematic ETFs and the benchmark. However, thematic ETFs exhibit higher Jenson’s α and Treynor ratio compared to the benchmark, while Sharpe ratio and β coefficient are inferior. Given that the majority of the studied thematic ETFs involve cross-border investments in technology, the third section conducts a sensitivity analysis. The results, altering the comparison group to the NASDAQ 100 index for the 26 cross-border investments, still show that, in the six and three months before establishment, the target index returns of thematic ETFs remained significantly higher than the comparison group (5.44% and 4.42%), cross-border investments ETFs or technology-focused ETFs. Generally, thematic ETFs tend to be established after the peak of the economic cycle for the respective theme, making it challenging to generate returns and performance exceeding the overall market. Except for a few thematic ETFs, the returns and performance indicators of most thematic ETFs fail to consistently outperform the benchmark. Moreover, as the investment period lengthens, the returns deteriorate, making them unsuitable as core asset allocations.