市值加權的投資組合方式一直廣為投資者所接受，其誘人之處莫過於集合前幾大市值的公司，並依據市值加權分配權重。在Markowitz(1952)提出的投資理論基礎下，討論了能夠在最小風險下獲得較高的預期報酬，應用在資產配置上即為最小變異數投資組合的投資策略。此觀念出現時，造成了市值加權投資組合其投資效率性受到質疑；由於影響市場變動的因子複雜性高，報酬率並不容易預測，因此出現了不少以最小風險為主要投資目標，而非追求高的預期報酬為主要目的之投資觀點。 本研究以市值加權投資組合為比較對象，探討最小變異數投資組合是否為更有效率的投資組合。以台灣50為市值加權投資組合，運用台灣50資料以滾動視窗的方式估計出最小變異數之權重，以此為最小變異數投資組合，比較兩者間的投資報酬率、績效表現。此外，本研究也將同時探討3種滾動視窗的期間長短，配合4種權重持有時間策略，以期了解參考資訊的時間長短、以及持有權重長短的策略是否會影響最小變異數投資組合的報酬率表現；最後，再運用迴歸分析，了解最小變異數投資組合與Fama-French三因子和動能因子之間的關聯性。 研究結果指出，當參考資訊越長期的滾動視窗估計方式，最小變異數投資組合會有較穩定的投資報酬率；若搭配長期的權重持有期間，能運用在長期投資。欲投資短期則可運用滾動視窗較短的估計方式，其在更新資訊較迅速的優勢之下，擁有較極端的報酬率的機率大，同時績效表現也較佳，此也代表最小變異數投資組合的權重估計若適當，則報酬率表現相當傑出；反之則有相當可觀的損失。觀察影響因子層面，最小變異數投資組合主要受到市場溢酬因子的影響，但根據短期權重持有的策略來說，則是動能因子影響越趨顯著。

近年來金融市場情勢瞬息萬變且波動劇烈而令投資人難以捉摸，導致被動式投資重獲投資人的青睞。市值加權法是最普遍被使用的指數編製方法，傳統市值加權指數的缺點，主要有投資組合主要集中於特定成份股與暴露於不會吸引風險溢價的各類風險因子中，這些缺點促成Smart Beta策略的發展，未來將有許多具備不同風險與報酬水準的指數供投資人參考。本研究選用近幾年提出不需仰賴預期報酬假說的等量風險貢獻度投資組合(ERC)建構方式，以台灣50為基準，利用其成份股建構投資組合，探討權重與風險分散特性，並且檢視其績效表現與報酬風險輪廓。為了有比較上的基礎，除了與台灣50做比較外，也另外選用以風險分散角度所配置的投資組合建構方法：等權重投資組合(EW)作對照。 　　研究結果發現全樣本時期，等量風險貢獻度投資組合在事後相較於台灣50擁有較低的波動度。當市場趨勢向下時，除了能維持其低波動的特性，還能提供某種程度上的抗跌能力。然而，以報酬率和Sharpe Ratio指標來看，表現皆不如台灣50，但優於等權重投資組合。同時，依不同經濟狀況與時間區間檢視投資組合績效表現，等量風險貢獻度投資組合能將波動度控制在較低水準，但績效表現上較不穩定。 　　本文透過HHI指數及吉尼係數衡量持股集中度與風險集中度。以持股權重分配的特性來說，台灣50極端集中於少數股票上，等量風險貢獻度投資組合和等權重投資組合則相對較平均。以風險權重分配的特性來說，台灣50風險權重過於集中，風險權重較平均分散的組合則為等量風險貢獻度投資組合。 / Traditional capitalization-weighted approaches are the most common ways to construct indexes. However, during market up turns, the capitalization-weighted indexes may be influenced by a small number of large-cap stocks. Smart beta indexes have recently prompted great interest among academic researchers and market practitioners. Risk-based indexes are an important category of smart beta. In this article, we explore equal risk contribution portfolio (ERC), which is risk-parity based smart beta. The portfolio implies to determine the weights so as to obtain the same risk contribution for each asset. The aim is to minimize the concentration in terms of risk contributions. In this paper, we examine whether or not ERC portfolio can outperform a buy and hold, capitalization-weighted and equally-weighted allocation in different economic environments. We also compute the parity in portfolio “risk allocation” and parity in “asset class allocation” using HHI index and Gini coefficient. 　　We consider here real-life applications with stock universe: the Yuanta /P-shares Taiwan Top 50 ETF. We compute smart beta portfolios by using the one-year empirical covariance matrix of stock returns. Empirical applications show that ERC portfolios generally are inferior in terms of return performance and Sharpe ratios. It does have some characteristics such as balanced risk allocation and less volatile performance characteristics. It also exposes to lower maximum drawdown. ERC portfolio provides the best ex ante and ex post “parity “in asset class risk contribution. On the other hand, the capitalization-weighted portfolio is concentrated in terms of weights and risk contributions. The statistics suggest to us that the capitalization-weighted portfolio’s superior Sharpe ratio is largely due to its higher returns.

過去實證研究顯示，匯率轉嫁程度並非一成不變，而是具有隨時間變動的特性。因此，有別於過去文獻大多採用滾動相關係數，本文則是使用Engle(2002)提出的動態條件相關係數模型，估計台灣於1982年至2014年間匯率變動與進口價格變動間的動態條件相關係數；並以其做為匯率轉嫁的代理變數，進而探討台灣匯率轉嫁的時間變動趨勢。我們的實證結果顯示，不論是用滾動相關係數還是動態條件相關係數，台灣的匯率轉嫁都明顯具有隨時間變動的特性。雖然5年期與10年期的滾動相關係數均在1997年前後分別呈現上升與下降的趨勢，動態條件相關係數則無類似的現象。然而，由於滾動相關係數容易受到滾動視窗樣本大小或滾動視窗有無包含極端值的影響，使得此方法較無法看出匯率轉嫁變動的準確時間點，而動態條件相關係數模型則可避免此問題。此外，本文實證發現，通膨環境與匯率波動是造成台灣匯率轉嫁隨時間變動的主要因子，對匯率轉嫁皆有顯著的正向影響。在排除1986年匯率轉嫁與進口滲透率呈現短暫負向關係的資料後，進口滲透率與匯率轉嫁的正向關係變為顯著，而進口滲透率也成為影響匯率轉嫁的原因之一。 / According to past empirical studies, it is believed that exchange rate pass -through (ERPT) has the time-varying nature. In this paper, we apply the Dynamic Conditional Correlation (DCC) model of Engle (2002), rather than the rolling correlation coefficient prevalently used by other studies, to analyze the time trend of ERPT for Taiwan. We estimate the dynamic condition correlation between the changes of exchange rate and the changes of import price using monthly data from 1982 to 2014 and use this correlation as a proxy for the degree of ERPT. Our empirical results show that ERPT for Taiwan, whether measured by the DCC or the rolling correlation coefficient, has a significant time- varying nature. In addition, both 5-year and 10-year window rolling correlation coefficient increase before 1997 and decline after 1997, which does not show in the DCC. However, the rolling correlation coefficient does not provide precise timings in the changes in ERPT, because of the dependence on the size of windows and whether or not outliers exist in the window. In contrast, the DCC does not have this kind of problem. Another important empirical result of this paper is that the inflation environment and the exchange rate volatility are main factors which explain the time-varying ERPT, and both of them have positive relation with ERPT. Moreover, the import penetration becomes positively significant after excluding data which shows temporary negative impact of the import penetration on ERPT in 1986.