My dissertation consists of two essays on mutual funds. The first essay examines the role of extreme positive returns on future fund flows using maximum style-adjusted daily returns (hereafter MAX) over the previous month. My results suggest that there is a positive and significant relation between MAX and future fund flows. The results are robust to controls for fund performance, fund size, age, turnover, fund fees, volatility, and skewness of fund returns. Of particular interest, this relation exits only in retail funds. Moreover, MAX is persistent from one month to the next, but MAX-based investment strategies are associated with lower risk-adjusted returns than investors could have achieved in otherwise similar funds. Overall, my analysis suggests that mutual fund investors are attracted to maximum style-adjusted daily returns, which is in line with the theoretical argument that investors exhibit a preference for lottery-like payoffs. These investors are successful in achieving a lottery-like return profile, but this strategy is costly in terms of expected returns The second essay studies the effect of recent and long-term mutual fund performance on future fund flows. I document that investors' response to recent performance depends on average long-term performance. In particular, a recent loser fund experiences outflows only if its longer-term performance is also poor. Similarly, recent good performance leads to more inflows only if the fund has also good long-run performance. In contrast, investors ignore recent performance if it provides a signal that conflicts with the longer-term signal. This implies that good fund managers with a longer-term focus will find it easier to attract future inflows than managers with a short-term horizon.

Kosowski, Timmermann, Wermers, and White (2006) find that certain growth-oriented fund managers have substantial skill but do not stipulate the particular skills that they possess. I use novel style timing models to examine in detail the timing skills of 3,181 US equity mutual funds classified as having a growth investment objective by Standard & Poor’s, over the period from 1993 to 2006. To control for idiosyncratic variation in mutual fund returns, the bootstrap method of Kosowski et al. is used to analyze the significance of alpha and timing coefficient estimates. To exclude the possibility that the observed timing ability is due to good luck, synthetic funds are examined as in Busse (1999). The results indicate that growth-oriented fund managers who earn abnormal returns demonstrate substantial growth timing skill, i.e. successful timing activity across the value/growth continuum. This observed growth timing ability accounts for at least 45% of abnormal returns and is persistent; the top 10% of funds which demonstrate growth timing ability in the past three years also demonstrate the best growth timing ability in the following year. Successful growth timing is confined to those managers who invest primarily in growth stocks. However, there is little evidence of successful market timing (i.e. forecasting future market states and weighting equity exposure accordingly), size timing (i.e. adjusting exposure between small and large capitalization stocks) or momentum timing (i.e. switching between momentum investing and contrarian investing strategies). The models employed clearly distinguish between growth timing and market timing skills, thereby avoiding a common misidentification problem.

As personal backgrounds and experiences vary, emotions stemming from exposure to violence shape a manager's risk perception and investment strategies. We document significant variation in the risk exposure of managers who were raised in states with higher per capita violence rates than those who were not. Although managers exposed to violence tend to hold more stocks in their portfolios, take less idiosyncratic risk, hold portfolios with betas closer to 1, and have less concentrated portfolios, these managers' risk-adjusted performance is not statistically different than that of their counterparts who were not exposed to violence.

Submitted by Vítor Uchôa Nunes (vitor.nunes@btgpactual.com) on 2011-07-12T19:23:36Z No. of bitstreams: 1 Tese (homolog - final) - Estudo da Sensibilidade do Fluxo de Investimento em Fundos Multimercado.pdf: 274894 bytes, checksum: 644c7fc5c8fd9522e335233e9cdab021 (MD5) / Approved for entry into archive by Vitor Souza (vitor.souza@fgv.br) on 2011-07-12T19:25:22Z (GMT) No. of bitstreams: 1 Tese (homolog - final) - Estudo da Sensibilidade do Fluxo de Investimento em Fundos Multimercado.pdf: 274894 bytes, checksum: 644c7fc5c8fd9522e335233e9cdab021 (MD5) / Made available in DSpace on 2011-08-03T18:32:48Z (GMT). No. of bitstreams: 1 Tese (homolog - final) - Estudo da Sensibilidade do Fluxo de Investimento em Fundos Multimercado.pdf: 274894 bytes, checksum: 644c7fc5c8fd9522e335233e9cdab021 (MD5) Previous issue date: 2011-04-18 / Este trabalho estuda o impacto causado no fluxo de investimento em fundos multimercado por variáveis como o retorno dos fundos, o retorno do benchmark, a volatilidade de mercado, o fluxo de investimento de estrangeiros na Bolsa de Valores do Estado de São Paulo e o desempenho dos fundos ajustados pelo risco. / This paper studies the impact caused on the investment flow of mutual funds by variables such as the return of funds, the benchmark return, the market volatility, the foreigner investment flow on the São Paulo Stock Exchange and the risk-adjusted performance of the funds.

This dissertation comprises two different financial essays. Essay 1, “An Applied Credit Score Model,” uses data from local credit union to predict the probability of default. Due to recent financial crisis regulation has been enacted that makes it essential to develop a probability of default model that will mitigate charge-off losses. Using discriminant analysis and logistic regression this paper will attempt to see how well credit score can predict probability of default. While credit score does an adequate job at classifying loans, misclassification of loans can be costly. Thus while credit score is a predictor, there is danger in relying solely on its information. Thus other variables are needed in order to more accurately be able to find the probability of default. Essay 2, “Christian Mutual Fund Performance,” draws attention to a much ignored type of funds, Christian mutual funds. The following questions are asked: How does Christian mutual fund perform compared to the market? Is there a difference in performance during recessions as indicated by literature? Is Christian mutual fund performance different than SRI funds? How do Catholic and Protestant fund perform? Looking at qualitative evidence, Christian mutual funds place much more importance on moral issue than SRI funds. Thus there is a clear difference in objectives and the type of screening that these two mutual fund pursue. Overall data reflects that screened data perform worse than the market, however during recession screened funds perform as well and at times better than the market. Christian mutual funds tends to perform worse than SRI funds.

O presente trabalho teve como objetivo principal identificar quais os fatores determinantes que afetam o desempenho dos Fundos de Previdência Renda Fixa na indústria brasileira de fundos, no período de janeiro 2005 a dezembro de 2011. A pesquisa em Fundos de Previdência aberta justifica-se na categoria Renda Fixa, pelos poucos estudos publicados, pelo crescimento contínuo da indústria de previdência aberta ao longo dos anos e pela ampliação das discussões sobre análise de desempenho e estilo de gestão nos fundos de Previdência Renda Fixa. A metodologia adotada identificou diferenças estatísticas significantes entre o desempenho dos fundos de Previdência Renda Fixa e os Fundos de Renda Fixa tradicionais, medidos nesse trabalho por meio de indicadores de desempenho e análise de estilo de gestão. A avaliação do desempenho dos fundos foi realizada por meio da aplicação de modelos quantitativos clássicos: Índice de Sharpe (1966) e Índice de Modigliani (1997). Posteriormente, para determinar os fatores de risco das carteiras dos fundos, foi utilizado o modelo proposto por Sharpe (1992), que ficou conhecido como Análise de Estilo Baseada no Retorno. Os resultados obtidos demonstram que os fundos de Previdência no período tiveram seus retornos abaixo dos retornos dos fundos de Renda fixa. A análise de estilo mostrou que os fundos de Previdência não concentram seus ativos vinculados à inflação, concentram-se mais em ativos financeiros atrelados à taxa de juros Selic. O desempenho dos Fundos de Previdência Renda Fixa indicou que, em todo o período estudado, os retornos dos fundos ficaram abaixo da taxa livre de risco, representada nesta pesquisa por 96% da taxa Selic. Revelou ainda que os fundos de previdência no período tiveram seus retornos abaixo dos retornos dos fundos de Renda Fixa. Foi realizado um estudo complementar sobre taxa de administração. Para os fundos de Renda Fixa, as taxas de administração foram reduzidas, no período, em média, para menos de 50%, enquanto os fundos de Previdência tiveram no mesmo período uma queda em torno de 20%. O retorno médio mensal dos fundos de Previdência que ficaram abaixo dos retornos dos fundos de Renda Fixa pode ser explicado, em parte, pela maior taxa de administração praticada. / This paper analyzes the performance of investment funds in Brazil, that are classified in the category of Pensions Provision Funds - Fixed Income Strategy (PPF-FIS), during the period of January 2005 to December 2011. There is scarce literature on this sub-field in Brazil, and the importance of this research is related to growing importance of PPF-FIS over the last years, and also to be a pioneer study in the analysis of performance and management style of them. The adopted methodology identified significant differences between the performance of traditional Fixed Income Funds and those PPF-FIS by means of performance indicators, and management style. Performance evaluation was done through the application of standard classical models like the Sharpe Ratio Index (1966), Modigliani Index (1997), and Return Based Style. The results showed that average returns of the PPF-FIS were below the average returns of traditional Fixed Income Funds. The style analysis of PPF-FIS´s portfolios showed them not concentrated on assets linked to inflation, but mostly to financial assets linked to the Selic Rate. The performance throughout the period was below the risk-free rate (measured as 96% of the Selic Rate), and in summary, the results show that PPF-FIS performed below the traditional Fixed Income Funds in terms of return. Other important achievement is related to administration fees that were reduced during the period by 50% in traditional Fixed Income Funds and by 20% in PPF-FIS. Higher administration fees are closely related to the poor performance of PPF-FIS compared to the traditional Fixed Income funds.

In the first essay, I create a hedge fund informed trading measure (ITM) that separates information related trades from liquidity driven trades. The results indicate that ITM predicts future stock returns at the trade level, thus is associated with information. By aggregating the most informed trades at the stock level, I find that stocks heavily purchased by informed hedge funds earn a significant alpha. The results indicate that the ITM performs better than some previously documented measures and is robust to two different versions of the measure. The second essay exploits the expiring nature of hedge fund lockups to create a new, within-fund proxy of funding liquidity risk. When funds have lower funding liquidity risk, risk-adjusted performance improves and exposure to tail risk increases. We use fund fixed-effect, a placebo approach, and a regression discontinuity design to establish a link between funding liquidity risk and the ability of funds to capitalize on risky mispricing. The third essay explores hedge fund managers ability to identify and trade on stock mispricing opportunity. We refer to the amount of capital that are is locked up and refrained from redemption as the stable capital, and study how it affects stock mispricing. We find that when funds have more lockup capital, they are more likely to take mispricing risks. Taking all funds together, more stable capital in the industry is driving the reduction or even correction of market-wide stock mispricing. Underpriced stocks benefit more than overpriced stock from hedge funds stable capital.

共同基金績效評量以夏普比率(Sharpe Ratio)最常被使用，但是由於夏普比率建構於常態分配的假設上，當基金報酬率不為常態時就可能產生偏誤。本文針對國內共同基金進行常態性檢定，發現基金報酬率分配呈現左偏、高狹峰的特質，並非常態分配，因此本文擷取風險值(VaR)衡量下方風險、又不需假設報酬率為常態分配的特長，將風險值應用在共同基金績效衡量上，以改善夏普比率在報酬率非常態分配下的偏誤，作為基金績效評估時輔助參考之用，並以國內共同資料進行實證研究，結論歸納如下： 共同基金績效排名衡量上，以風險值取代夏普比率標準差的指標所得到的排名會與夏普比率所得到的排名的確有所差異。一般類股票型基金方面，以風險值取代夏普比率標準差的指標排名相對夏普比率提升的基金都具有風險值較小的特點；另一方面，上櫃股票型基金及科技類股票型基金排名因報酬率差異較大，所以出現報酬率主導排名順序，改變風險衡量方式影響排名不大的現象。 本文比較以風險值取代夏普比率標準差的指標及以標竿報酬率代替無風險利率的指標、以風險值取代夏普比率標準差的指標及報酬風險值均考慮市場影響的指標，瞭解所處市場走勢對基金績效的影響，實證結果發現上櫃型基金排名均往前攀升；科技類股票型基金在考慮市場因素後所獲得的排名評價有後退之現象產生。 在指標預測性方面，夏普比率和以風險值取代夏普比率標準差的指標在統計上不具顯著性；以標竿報酬率代替無風險利率的指標和報酬風險值均考慮市場影響的指標則在統計上具顯著性，具有預測參考價值。

none

Hedge fund managers are characterised as either market timers or asset pickers . Their superior performance can be attributed to either timing skill, selection ability or a combination of both. In the existing literature, average hedge fund performance across the entire time span under investigation is usually investigated and measured, and hence, potentially certain subtle but important features exhibited in different time periods can be averaged out in the analysis. This thesis investigates the structural breaks in the selection ability and timing skill of hedge fund managers. This research issue is of particular importance when the hedge fund performance before, during and after the recent financial crisis is compared and contrasted. This thesis conducts a structural break analysis of hedge fund managers performance in relation to market-wide liquidity and liquidity commonality in the foreign exchange (FX) market. Liquidity commonality captures the co-movement of individual asset liquidities. The measure adopted in the existing literature has several limitations. This thesis proposes a new measure, termed the Beta Index, which is derived from the time-varying exposure of individual liquidities to market liquidity movements. It is shown that the developed Beta Index is more able to identify the level of liquidity commonality in the FX market. It is also more flexible in measuring commonality with different data sampling frequency. The obtained empirical results have some practical implications. They show that the selection skill and timing ability of hedge fund managers are subject to regime switches. Under severe market conditions, most hedge fund managers possess the skill to time FX market-wide liquidity and are able to reduce losses from the FX market by reducing their funds FX exposure prior to the FX market-wide liquidity deteriorations. In the meantime, most hedge fund managers are able to deliver excess returns from time to time due to their selection ability. However, when sudden shocks of crisis occur, they fail to forecast the unexpected behaviour in the price of individual assets underlying the funds and display unsuccessful selection ability. In addition, the results suggest that many hedge funds are exposed to the FX liquidity commonality risk which impairs hedging strategies and diversification performance.