Türkiye’deki yatırım fonu performansına aktif benzer karşılaştırma ölçütlerine dayalı olarak yakından bir bakış.


Tezin Türü: Yüksek Lisans

Tezin Yürütüldüğü Kurum: Orta Doğu Teknik Üniversitesi, İktisadi ve İdari Bilimler Fakültesi, İşletme Bölümü, Türkiye

Tezin Onay Tarihi: 2017

Tezin Dili: İngilizce

Öğrenci: Aybars Furkan Şanap

Danışman: HALİT ENGİN KÜÇÜKKAYA

Özet:

This master's thesis aims to evaluate mutual fund performance in Turkey on a risk-adjusted basis with various approaches. First, commonly used Capital Asset Pricing Model (CAPM) and Carhart Four-Factor Model are employed for performance analysis. Then, active peer benchmarks (APB) are utilized to account for market-capitalization characteristics of portfolio stockholdings and to control commonalities in unsystematic risk-taking of different mutual funds. The empirical results are mixed for the sampled fifty-two equity mutual funds over the period between July 2012 and June 2016. CAPM provides that only two funds have abnormal performance while none of the funds generate a significant positive alpha with Carhart Four-Factor Model. APBs which are the average excess returns of funds following the same market cap-strategy are further used to isolate fund-wise unsystematic risk-taking and selectivity skill. The first stage of the model with only APB residuals incorporated as a fifth factor reveals that the sampled funds cannot generate excess return even after controlling the common unsystematic risk-taking. However, when the alpha associated with the common unsystematic risk taking of funds in the same APB group is also controlled, it is observed that two funds generate significant positive alphas. On the other hand, APB-Augmented Model performs better than generic models in terms of capturing common unsystematic risk-taking. This study documents that four-factor models decreases the percentage of significant positive pairwise residual correlations only by one-tenth of its previous level in CAPM. Nevertheless, APB-Augmented Model provides a reduction of approximately 50% compared to four-factor model.