Are the ratings useful tools selecting mutual funds?

UNIVERSITY*OF*SANTIAGO*DE*COMPOSTELA** DEPARTMENT*OF*FINANCIAL*ECONOMICS*AND*ACCOUNTING** * * UNIVERSIDAD DE SANTIAGO DE COMPOSTELA DEPARTAMENTO DE ECONOMÍA FINANCIERA Y CONTABILIDAD * * * Are the ratings useful tools selecting mutual funds? ¿Los ratings son herramientas útiles en la selección de fondos de inversión? Renato*Heitor*Correia*Domingues Santiago(de(Compostela(2017( ( ! UNIVERSIDAD!DE!SANTIAGO!DE!COMPOSTELA! DEPARTAMENTO!DE!ECONOMÍA!FINANCIERA!Y!CONTABILIDAD! ! ! ! ! ! ! Are!the!ratings!useful!tools! selecting!mutual!funds?! ¿Los!ratings!son!herramientas!útiles!en!la!selección!de!fondos!de!inversión?! ! Tesis!para!la!obtención!del!grado!de!doctor,!presentada!por!el!licenciado!Don!Renato!Heitor! Correia! Domingues,! la! cual! fue! realizada! en! el! Departamento! de! Economía! Financiera! y! Contabilidad!de!la!Universidad!de!Santiago!de!Compostela,!bajo!la!dirección!de!Don!Luis!Alberto! Otero! González,! profesor! titular! de! Economía! Financiera! y! Contabilidad! y! Don! Pablo! Duran! Santomil,! profesor! ayudante! doctor! de! Organización! de! Empresas! y! Comercialización!de! la! Universidad!de!Santiago!de!Compostela.! ! Vº!Bº! ! ! ! ! ! ! D.!Luis!Alberto!Otero!González! ! ! ! !D.!Pablo!Duran!Santomil! D.! LUIS! ALBERTO! OTERO! GONZÁLEZ,! profesor! titular! del! Departamento! de! Economía! Financiera!y!Contabilidad!de!la!Universidad!de!Santiago!de!Compostela!y!D.!PABLO!DURAN! SANTOMIL,!profesor!ayudante!doctor!de!Organización!de!Empresas!y!Comercialización!de! la!Universidad!de!Santiago!de!Compostela! ! ! CERTIFICAMOS! que! D.! Renato! Heitor! Correia! Domingues! ha! realizado! bajo! nuestra! dirección!el!trabajo!de!investigación!“¿Los!ratings!son!herramientas!útiles!en!la!selección! de! fondos! de! inversión?”.! Este! trabajo! reúne! las! condiciones! necesarias! para! ser! presentado!y!juzgado!como!tesis!doctoral!y,!por!tanto,!para!optar!al!grado!de!doctor!en! Ciencias!Económicas!y!Empresariales.!! ! Santiago!de!Compostela,!2017! ! Fdo.!Prof.!Dr.!Luís!A.!Otero!González! Profesor!titular!del!Departamento!de! Economía!Financiera!y!Contabilidad! Universidad!de!Santiago!de!Compostela! Fdo.!Prof.!Dr.!Pablo!Durán!Santomil! Profesor!Ayudante!Doctor!del! Departamento!de!Organización!de! Empresas!y!Comercialización! Universidad!de!Santiago!de!Compostela! For$my$family,$especially$for$Renata! Acknowledgement 1 Acknowledgement My journey through this thesis was long and it would not be possible without the support of several people. First and foremost, I would like to express my gratitude to Luis Alberto Otero González for his fantastic supervision. His encouragement, patience and guidance were vital for my development as a researcher. To Pablo Duran Santomil, for his co-supervision that helped me to improve time management, technical details and quality standards. It is hard to find the words to express how grateful I feel for their work in this project. To Nuno Miguel Cruz Domingues for encouraging me to go forward with this PhD, for changing his critical opinion and for participating in this study's debate.To all my dear family for their outstanding support and endless encouragement that helped to nurture a positive attitude throughout the challenges. To Marta Venade, for her friendship and help with the English revision. Last but not least, a big thank you to all my PhD Professors and their fellow students for their availability. Thank you very much again, it has been a pleasure to work with you all. Are the ratings useful tools selecting mutual funds? 8 tamaño y valor. La muestra contiene 10.772 observaciones mensuales y un promedio de 220 fondos de inversión clasificados, con una buena representación en cada nivel excepto en el caso negativo. Siguiendo a Amstrong et al. (2016) distinguimos entre fondos "no recomendados" y "recomendados", donde los “recomendados” están compuestos por las categorías de oro a bronce, y neutral y negativo se clasifican en "no recomendado". En general, más del 80% de los fondos son clasificados como "recomendados", y esto es explicado por Morningstar por el hecho de que priorizan fondos de alta calidad y en general son más grandes y con menores costes y tasas de rotación. Para el análisis empírico, hemos estimado distintos modelos para evaluar el rendimiento en base los ratings Analyst y sus pilares. Dado que es un rating con carácter prospectivo (fordward looking) nuestra hipótesis de investigación es que reflejara el rendimiento futuro en el largo plazo, de forma los fondos recomendados obtendrían mayor rendimiento futuro. Se calcula que no se muestra el tiempo de retorno de la respuesta por 12 y 36 meses después de la inicialización de la calificación y, a continuación, devolver las diferentes métricas de performance utilizando oro, plata y bronce añadiendo después variables de control. Utilizamos lo método basado en los datos de panel que puede controlar los efectos individuales de los efectos de la colinealidad y la eficiencia, entre otros (Baltagi, 2010). A semejanza del capítulo anterior, los modelos de panel fueron inicialmente estimados sin variables de control para evaluar el efecto de seleccionar fondos basados exclusivamente en el rating Analyst. Posteriormente, se han incluido otras variables explicativas tales como costes, tamaño y años. En todos los casos se han incluido variables de control la categoría del fondo y el efecto temporal de la rentabilidad. Si bien los resultados dependen de la métrica utilizada, en general, sólo los fondos clasificados como Oro muestran un mejor desempeño en un horizonte de 12 meses sobre los fondos "no recomendados" en términos de la ratio de Sharpe. A 36 meses los resultados son más decepcionantes. Por lo tanto, nuestro análisis revela la incapacidad de las calificaciones de los analistas para identificar los fondos que superan a sus pares excepto en el caso de los fondos de oro, donde los resultados muestran un mejor desempeño que los clasificados como no recomendados a corto plazo. Las diferencias con estudios previos pueden deberse al hecho de considerar en el estudio una muestra diferente centrada exclusivamente en los EE.UU., la metodología de datos de panel y el período considerado. Posteriormente hemos incluido los pilares en los que los ratings Analyst se desglosan para analizar si el rendimiento futuro está relacionado con cualquier dimensión específica. El hecho de que haya muy pocos fondos de inversión con un pilar negativo conduce a una Resumen 9 comparación entre pilares fundamentalmente positivos y neutrales. Los resultados obtenidos muestran, en general, coeficientes no significativos, con signo negativo y positivo. La alta subjetividad que puede conducir al proceso de evaluación y la dificultad para establecer un umbral entre una evaluación positiva y otra neutral pueden explicar los resultados obtenidos. Para entender si la combinación de ambas calificaciones (cuantitativas y cualitativas) pueden ayudar en el proceso de identificación de los mejores resultados, también estudiamos el efecto de considerar conjuntamente los dos tipos de ratings. Así, se crearon diferentes variables que resultan de la combinación del mejor analista (Oro a Bronce) y estrellas (4 o 5 estrellas) y considerando el resto como "no recomendado. Los resultados del análisis conjunto de los fondos muestran que cuando las calificaciones se utilizan aisladamente, sólo los fondos de cuatro o cinco estrellas y los fondos Oro superan a la clasificación cuantitativa más baja a corto plazo. Cuando ambos criterios se combinan los fondos Oro y cuatro o cinco estrellas tienen un mejor desempeño a 12 meses, pero no a 36 meses. Esto significa que los inversores que basan sus decisiones en ambos criterios deben monitorear las carteras anualmente y verificar que continúan mantienen ambas calificaciones. Por otro lado, la combinación de las dos calificaciones tiene resultados a medio plazo, con un mayor rendimiento en términos de Sharpe para los fondos Bronce de 3, 4 o 5 estrella y Oro de tres estrellas, presentando sólo un signo negativo los fondos Plata de tres estrellas. Para dar más consistencia al estudio, hicimos tests complementarios de robustez. De esta forma utilizamos la regresión cuantil para extender el modelo de regresión a los cuantiles condicionales de las diferentes métricas de rendimiento, ya que está técnica puede ser más apropiada para un universo heterogéneo de fondos de inversión donde las estrategias y los objetivos pueden variar (Chen y Huang, 2011). Este modelo nos permite capturar información sobre los coeficientes en diferentes cuantiles de la variable dependiente dado el conjunto de variables endógenas. Los resultados de la regresión cuantil muestran que, en general, los signos no son significativos para la mayoría de las calificaciones Analyst, siendo los fondos Oro los únicos que superan a los “no recomendados” en algunos cuantiles. Cuando se incluye el efecto de los costes o el tamaño de los fondos, las calificaciones Oro que son significativas dejan de serlo. Por lo tanto, es razonable pensar que tanto el tamaño de los fondos como los costes pueden explicar las diferencias de desempeño más que las calificaciones de los analistas. En el Capítulo 4, nuestro estudio se centra en el nuevo rating de sostenibilidad de Morningstar (Morningstar Sustainability Rating), en concreto emplearemos los puntuaciones o scores de Sostenibilidad y de ESG (iniciales de Environmental, Social y Governance). El Are the ratings useful tools selecting mutual funds? 10 objetivo de este capítulo es extender el tradicional debate de si las inversiones en fondos de inversión sostenibles afectan a la performance de un forma positiva, neutra o negativa. Las inversiones socialmente responsables (SRI) en fondos de inversión es una opción que tiene en cuenta los criterios ambientales, sociales y de gobierno corporativo (ESG) para generar valor a largo plazo. Otros nombres asociados a este tipo de inversión en la literatura son el de inversión social, sostenible, socialmente consciente, verde, responsable o ética. Los tradicionales estudios realizados hasta la fecha han empleado una variable dicotómica para valorar si un fondo es socialmente responsable o no con base en cómo la gestora lo declara en su Folleto de Gestión. Sin embargo, la reciente publicación en el año 2016 de Morningstar de los ratings y puntuaciones de sostenibilidad nos permite analizar las diferencias, en vez de aplicando un criterio binario (si/no) las puntuaciones alcanzadas por cada fondo. Varios estudios han concluido que las empresas que tienen políticas y prácticas de responsabilidad social son buenas inversiones y, por lo tanto, los fondos que invierten en ellas. Recientemente, por ejemplo, Friede, Busch y Bassen (2015) llevó a cabo un meta análisis de unos 2.200 estudios empíricos, convirtiéndolo en la revisión más completa de la investigación académica sobre este tema. Encontraron que la mayoría de los estudios muestran una correlación positiva entre los factores ESG y el rendimiento financiero. Pero incluso a pesar de las investigaciones llevadas a cabo hasta la fecha todavía hay un debate acerca de si este tipo de inversiones pueden crear valor para los inversores o no. Aunque según Lewis y Mackenzie (2000) y Webley, Lewis y Mackenzie (2001) algunos inversores en fondos SRI están dispuestos a aceptar rendimientos menores por su postura moral, el desempeño de los fondos SRI y convencionales es una cuestión todavía abierta, especialmente cuando se analizan las puntuaciones de sostenibilidad de un fondo, ya que hasta la fecha sólo ha sido analizado por El Ghoul y Karoui (2017). En la literatura que realiza estudios con variables dicotómicas para diferenciar los fondos SRI de los fondos convencionales se han encontrado evidencias contrapuestas de si éstos obtienen mejores resultados. Autores como Junkus y Berry (2015) sustentan que las inversiones en fondos SRI tienen desempeño muy similar que los fondos convencionales. Estudios como Luther, Mattako y Corner (1992) y Mallin, Saadouni y Briston (1995) apoyan la idea de que los fondos SRI superan a los índices del mercado. Pero la teoría más convencional es que los fondos mutuos de ISR tienen el mismo rendimiento que los fondos de otros, y autores como Hamilton, Jo y Statman (1993), Kreander, Gray, Power y Sinclair (2002, 2005), Gregory y Whitakker Bauer, Derwall y Otten (2007) o Humphrey, Warren y Boon (2016) están en línea con esta teoría. Nuestro estudio aporta como principal novedad a la actual literatura sobre la rentabilidad Resumen 11 de los fondos de inversión socialmente responsables el examen del efecto del grado de sostenibilidad, medido a través de las puntuaciones de sostenibilidad y de ESG de Morningstar. Las puntuaciones de sostenibilidad y ESG de Morningstar se elaboran para cada fondo que invierte al menos el 50% de sus activos en compañías con puntuaciones calculadas por Sustainalytics. Sustainalytics es una compañía líder en las valoraciones de criterios ESG de compañías de todo el mundo. La metodología de ESG de Sustainalytics consiste en distintos indicadores ESG, de 120 a 150 en función de la industria, para medir las prácticas sostenibles de una empresa. Sustainalytics evalúa el desempeño de la empresa en cada uno de estos indicadores a partir de varias fuentes de datos internas y externas (Sustainalytics, 2016). Estos indicadores se engloban en los tres pilares de la puntuación ESG, que son medio ambiente, social y gobernanza. En cada pilar se distinguen varias categorías de indicadores: básicos y específicos del sector. De esta forma para cada compañía obtiene mediante agregación de los indicadores una puntuación para cada uno de los pilares ESG, y por medio de las denominadas controversias, de la puntuación global de sostenibilidad. Morningstar agrega las puntuaciones de los activos que forman la cartera de un fondo de inversión de forma normalizada para obtener la puntuación total del fondo en cada pilar ESG, la puntuación total de ESG, las controversias del fondo, y finalmente, por diferencia entre la puntuación ESG y las controversias, la puntuación de sostenibilidad de un fondo de inversión, de forma que este puede obtener un valor máximo teórico de 100. Nuestro estudio tiene una muestra inicial de 1.593 fondos de inversión de renta variable europea de tipo abierto (open funds) con puntuaciones de sostenibilidad de Morningstar en noviembre de 2016. La selección de fondos se ha hecho evitando problemas de multicolinealidad, por lo que se ha selecciona sólo una clase. Para cada fondo se analiza el efecto que tienen las puntuaciones de sostenibilidad, ESG y de cada uno de los pilares de forma independiente en diferentes medidas de desempeño. Además, se ha tenido en cuenta otras variables como el tamaño, los gastos y la antigüedad del fondo. El número final de fondos varió cuando se consideran los costes, de forma que la muestra se reduce de 1.593 a 571 motivados por la falta de datos en Morningstar Direct para todas las variables consideradas. Hemos analizado el rendimiento y el efecto de riesgo utilizando las métricas de rentabilidad y riesgo de los últimos dos años basados en el trabajo de Wimmer (2012), que muestra que las puntuaciones ESG persisten durante dos años motivadas por los cambios en las tenencias de activos de los fondos de inversión SRI. En particular, como medidas de performance empleamos la rentabilidad anual, el alfa de Jensen y ratio de Sharpe. Además, se ha considerado el efecto que la sostenibilidad tiene en los flujos de efectivo de un fondo y también su efecto sobre el valor en riesgo (VaR). Are the ratings useful tools selecting mutual funds? 12 Los resultados obtenidos muestran que hay un gran número de fondos que no se declaran sostenibles, pero su cartera es comparable a los fondos de inversión que se declaran sostenibles. Además, el parámetro asociado en la regresión a la puntuación de sostenibilidad es significativo, explicando el nivel de desempeño de todas las métricas analizadas (alfa, Sharpe y rendimiento neto anual), con signo negativo en la mayoría de los modelos. Utilizando una variable dummy convencional para identificar los fondos de inversión SRI, los resultados son significativos, pero, al contrario, mostrando que considerar el nivel de sostenibilidad puede ayudar a entender mejor la relación entre desempeño y responsabilidad social. Nuestros resultados están de acuerdo con Statman y Glushkov (2016), quienes concluyen que la falta de criterios claramente definidos para distinguir los fondos SRI afecta a los resultados de los estudios previos de la literatura, y eso puede explicar el por qué se encuentran en la misma resultados contradictorios. Asimismo, obtuvimos resultados similares a los de El Ghoul y Karoui (2017) para el mercado de fondos de inversión de Estados Unidos. Utilizando los diferentes pilares de las puntuaciones de ESG (ambiental, social y de gobierno), también alcanzamos una relación negativa entre dichas dimensiones y el rendimiento, demostrando que todas las dimensiones desempeñan un papel importante en la explicación del desempeño financiero. En términos de riesgo, el nivel de sostenibilidad está relacionado negativa y significativamente con el VaR del fondo, apoyando que los fondos de inversión mejor calificados se comportan mejor frente a las pérdidas extremas del mercado. El signo contrario se encuentra en el caso del parámetro asociado a la variable dummy convencional, mostrando las ventajas de emplear una medida cuantitativa de sostenibilidad para evaluar el riesgo de los activos. Este resultado podría significar que los gestores de fondos de inversión SRI basan sus decisiones en un análisis más profundo que resulta en una reducción significativa en el riesgo de sus decisiones de inversión. Nuestra evidencia de trabajo que la puntuación de sostenibilidad puede ser utilizado por los inversores preocupados por las pérdidas extremas y no sólo por los inversores motivados por los valores de sostenibilidad. Realizamos pruebas de robustez adicionales para verificar la consistencia de nuestros resultados y para proporcionar otros análisis complementarios. Hemos realizado una regresión cuantílica para analizar si los anteriores efectos difieren en los diferentes cuantiles analizados, lo que indicaría diferencias en las habilidades de los gestores respecto el desempeño, no obteniendo diferencias con los resultados previos. Además, se han recalculado los modelos excluyendo del análisis la variable gastos del fondo, dado que dicha variable reducía la muestra, no obteniendo resultados diferentes. Por último, analizando el efecto de la sostenibilidad en los flujos de los fondos se concluye que las rentabilidades no ajustadas al riesgo tienen mayor influencia en las Resumen 13 decisiones de inversión. El score de sostenibilidad afecta de una forma significativa a los flujos de los fondos, por lo que los fondos de mayor calificación tuvieron un mayor volumen de inversión, siendo también significativo el efecto de la variable ficticia de sostenibilidad. Index 15 Index ABBREVIATIONS -------------------------------------------------------------------------------------------------- 17 TABLES ---------------------------------------------------------------------------------------------------------------- 19 FIGURES -------------------------------------------------------------------------------------------------------------- 21 CHAPTER I: INTRODUCTION --------------------------------------------------------------------------------- 23 1. MOTIVATION OF THE STUDY --------------------------------------------------------------------------------- 25 2. CONTRIBUTION OF THE STUDY ------------------------------------------------------------------------------ 26 3. STRUCTURE OF THE STUDY ---------------------------------------------------------------------------------- 27 4. OBJECTIVES ---------------------------------------------------------------------------------------------------- 28 5. HYPOTHESES TESTING ---------------------------------------------------------------------------------------- 28 6. METHODOLOGY ----------------------------------------------------------------------------------------------- 30 CHAPTER II: ARE QUANTITATIVE RATINGS USEFUL TOOLS SELECTING MUTUAL FUNDS? ---------------------------------------------------------------------------------------------------------------- 33 1. INTRODUCTION ------------------------------------------------------------------------------------------------ 33 2. BACKGROUND MORNINGSTAR STAR RATINGS ----------------------------------------------------------- 34 3. LITERATURE REVIEW ----------------------------------------------------------------------------------------- 35 4. EMPIRICAL STUDY --------------------------------------------------------------------------------------------- 41 4.1. Performance Metrics --------------------------------------------------------------------------------- 41 4.2. Models -------------------------------------------------------------------------------------------------- 43 4.3. Results for rating models ---------------------------------------------------------------------------- 45 4.4. Results with net expenses and other variables ---------------------------------------------------- 50 4.5. Results with downside risk --------------------------------------------------------------------------- 52 5. ROBUSTNESS --------------------------------------------------------------------------------------------------- 55 6. CONCLUSIONS ------------------------------------------------------------------------------------------------- 55 CHAPTER III: DOES MORNINGSTAR ANALYST RATING MATTERS FOR MUTUAL FUNDS? ---------------------------------------------------------------------------------------------------------------- 57 1. INTRODUCTION ------------------------------------------------------------------------------------------------ 57 2. MORNINGSTAR QUALITATIVE RATINGS. ------------------------------------------------------------------- 59 2.1. Stewardship Grade ----------------------------------------------------------------------------------- 59 2.2. Morningstar Analyst Rating ------------------------------------------------------------------------- 61 3. PREVIOUS RESEARCH. ---------------------------------------------------------------------------------------- 62 4. EMPIRICAL STUDY --------------------------------------------------------------------------------------------- 69 4.1. Data and sample -------------------------------------------------------------------------------------- 69 Are the ratings useful tools selecting mutual funds? 16 4.2. Models -------------------------------------------------------------------------------------------------- 71 4.3. Results for Analyst ratings (forward looking) ----------------------------------------------------- 72 4.4. Results for some Pillars included in the Analyst ratings (forward looking) ------------------ 75 4.5. Results for stars ratings (backward looking) ------------------------------------------------------ 75 4.6. Results for best stars and analyst ratings (combining forward and backward looking) ---- 76 5. ROBUSTNESS --------------------------------------------------------------------------------------------------- 77 6. CONCLUSIONS ------------------------------------------------------------------------------------------------- 80 CHAPTER IV: DOES SUSTAINABILITY SCORE IMPACT MUTUAL FUND PERFORMANCE? -------------------------------------------------------------------------------------------------- 82 1. INTRODUCTION ------------------------------------------------------------------------------------------------ 82 2. LITERATURE REVIEW ----------------------------------------------------------------------------------------- 86 3. BACKGROUND OF SUSTAINALYTICS´ METHODOLOGY AND MORNINGSTAR -------------------------- 93 3.1 Sustainability Scores. --------------------------------------------------------------------------------- 93 4. EMPIRICAL STUDY --------------------------------------------------------------------------------------------- 96 3.1. Sample -------------------------------------------------------------------------------------------------- 96 3.2. Variables construction ------------------------------------------------------------------------------- 97 3.3. Performance variables ------------------------------------------------------------------------------- 98 3.4. Downside risk variables ------------------------------------------------------------------------------ 99 3.5. Flow of funds ------------------------------------------------------------------------------------------ 99 3.6. Descriptive statistics -------------------------------------------------------------------------------- 100 3.7. Fund Performance and Sustainability Scores --------------------------------------------------- 101 3.8. Downside risk and sustainability scores --------------------------------------------------------- 104 3.9. Flows and sustainability scores ------------------------------------------------------------------- 106 5. ROBUSTNESS ------------------------------------------------------------------------------------------------- 107 6. CONCLUSION ------------------------------------------------------------------------------------------------- 111 CHAPTER V. CONCLUSIONS, LIMITATIONS AND FUTURE RESEARCH -------------------- 113 REFERENCES ----------------------------------------------------------------------------------------------------- 118 APENDDIX ---------------------------------------------------------------------------------------------------------- 127 Abbreviations 17 Abbreviations CFAChartered Financial Analyst ETFsExchanged Traded Funds ESG -Environmental, Social and Governance L1One lag L3 -Tree lag MSCIMorgan Stanley Capital International MSCI ESGMorgan Stanley Capital International Environmental, Social and Governance MRAR-Morningstar Risk-Adjusted Return NGOsNon-Governmental Organizations OLS -Ordinary Least Squares Q75Quintile 75% SR- -Social Responsibility SRI -Social Responsibility Investment VaR -Value at Risk Are the ratings useful tools selecting mutual funds? 24 some extent, future performances. Authors such as Morey and Gotesman (2006), Müller and Weber (2014) and Meinhardt (2014) have demonstrated through their studies that quantitative ratings may be able to predict the future performance of mutual funds. Star ratings are the most popular among mutual fund investors, because ratings are goals based on historical performance and easy to understand. However, Morningstar has also developed other ratings that further enhance the quality aspect and can be useful, and even complement quantitative ratings, by helping to understand whether ratings can contribute to the selection of the best funds. Among the most popular are the Star Ratings or quantitative ratings, but also the qualitative Analyst ratings and the Stewardship Grades, including the recent Morningstar's Sustainability Rating. The selection of funds based exclusively on historical performance or quantitative ratings, excludes a set of qualitative factors that can explain future performance. This is why, in addition to quantitative ratings it has appeared that ratings based on analyst opinions evaluate aspects of mutual funds as: Governance, Process, People, Parent, Board Quality, Corporate Culture, Fees, Manager Incentives or Regulatory Issues. The existence of several studies about quantitative and qualitative ratings are not as popular since there are only a very limited number of researches that focus on the the ability to select good funds based on qualitative ratings. In the particular case of Morningstar, there are two alternatives: Analyst Ratings and Stewardship Grade. In our study of qualitative ratings, we will focus on Analyst ratings as it is the less discussed subject in literature, since there are more recent studies like Morningstar who launched them in 2011. Analysts ratings rate the funds based on their belief in the fund's ability to outperform its benchmark or its competitors in the long run. To reach a rating, analysts evaluate five key pillars that our experience has shown as critical to a fund's ability to succeed like: Personal, Management Firm, Process, Performance, and Price. In relation to the star ratings there are some studies trying to understand the role of these in the selection of mutual funds. Regarding qualitative ratings, there are not many studies, since they are relatively recent. Wellman and Zhou (2007), Seng (2009) and Gottesmann and Morey (2012) studied the Stewardship Grades, and Kamal (2013) and Armstrong, Genc y Verbeek (2016) are the unique authors that studied the predictive power of Analyst ratings. In general, their results have shown the ability to predict future out-of-sample performance: higher Analyst Ratings do predict better future performance and it is verified in funds with better rating (Gold medal). Finally, we investigate the effect of Morningstar's Sustainability and ESG scores on performance, flows and risk. These scores are compiled for each fund that invests at least 50% of its assets in companies rated by Sustainalytics (a leading company in the ESG criteria valuations Chapter I: Introduction 25 of companies worldwide). It’s methodology consists of different ESG indicators between 120 and 150, that depending on the industry, measure the sustainable the companies practices. Sustainalytics evaluates the company's performance in each of these indicators based on several internal and external data sources (Sustainalytics, 2016), it uses two types of indicator templates: basic and sector-specific. Each company obtains a score by aggregating the indicators for each of the ESG pillars, the global ESG score and the controversies of the company. Morningstar adds the asset scores that form the portfolio of an investment fund in a standardized way to obtain the total score of the fund in each ESG pillar, total ESG score, fund controversies, and finally, by difference between the score ESG and the controversies, the sustainability score of an investment fund. Although according to Lewis and Mackenzie (2000) and Webley, Lewis and Mackenzie (2001) some investors in socially responsible investment (SRI) funds are willing to accept lower returns for their moral stance making the performance of SRI funds and conventional funds an open question. As Junkus and Berry (2015) sustain, after a review of the most recent work in major finance journals on SRI: “the performance of SR mutual funds and indexes are not generally significantly different to conventional funds or indexes, but again these results are also highly dependent on model specification, time period, benchmark, and other characteristics of the study”. As far as we know, only El Ghoul and Karoui (2017) use a scores to study the effect on sustainability on fund performance and flows, concluding that higher values displaying poorer performances and weaker performance-flow relations. 1. Motivation of the study The motivation behind this study lies in the diffusion of the use of funds´ ratings in the making of their investment decisions. The Morningstar´s ratings are the most important in the industry of mutual funds because it is a very respected investment research firm among investors, sustained by many studies to prove that investors care about the Ratings such as: Faff, Parwada and Poh (2007) and Del Guercio and Tkac (2008), among others that find quantitative ratings important for the flows of mutual funds. This raises the question of whether investors are well advised to pay attention to Morningstar ratings. There are many funds, with different strategies and different assets so it is impossible to know and control all the information about a huge amount of funds and assets with many specification. Only the specialists of the various branches of investment have enough knowledge to do it, and it is impossible for non-institutional investor to know all the investment strategies and understand them. Are the ratings useful tools selecting mutual funds? 26 Persistence plays a very important role on the literature debate that allows us to see if there is the ability of certain managers to achieve a better performance by beating the market in a consistent way. So, it will be important to see if investors can make their investment decisions and to realize if extent past performance may contain information about future performance. The question whether qualitative aspects in management are more determinant or can complement the quantitative information, as is the case of historical returns, will be another important aspect to realize in this thesis. Finally, this work is also of interest, because Morningstar is following the needs of investors and in the most current debates. An example of this is not only the qualitative aspects that were a clear response to Morningstar's new investor needs but also the creation of a rating that evaluates SRI behaviors of mutual funds so it is an important motivation to realize to what extent SRI affects performance. Finally, choosing the best funds with the information contained in the ratings can be a very efficient and simple way for investors, since the ratings greatly simplify all the information contained in the investments . 2. Contribution of the study This study intends to give several contributions to the academic literature, as well as to help in the decision making of the investors and institutions. There are many studies that focus on the quantitative ratings of mutual funds and the persistence of returns to see if there may be information on past returns to predict future returns. However, there are not many known studies that focus on the predictive power of Morningstar Star Ratings. Some studies such as Blake and Morey (2000), Morey (2002, 2005) and some others, studied the predictive power of quantitative ratings obtaining weak results, but some recent studies have found that ratings may lead to the choice of the best funds that ratings can predict, in some extent, future performances. Authors such as Morey and Gotesman (2006), Müller and Weber (2014) and Meinhardt (2014) have demonstrated through their studies that ratings may be able to predict the future performance of mutual funds, but none of them used a database with so many funds and with such a long-time spectrum like as ours. There are few studies that have focused on the predictive power of star ratings with databases that have a significant share of funds with ratings but after the change in methodology we do not know any of them. In this empirical analysis, control variables such as Net Expenses, Turnover, among others, are also added to see if they can help to increase the predictive power of Chapter I: Introduction 27 star rating when used together, or to also understand if these control variables can explain the star ratings and past performance. In the following empirical analysis, we focus on Analyst ratings. Kamal (2013) and Armstrong, Genc y Verbeek (2016) are the unique authors that studied the predictive power of Analyst ratings, but no one has studied Analyst ratings and their predictive power in an exhaustive way. The existing studies use a database with few funds and the time horizon is very short, since Morningstar's analyst ratings are relatively recent. In this study we develop a rating in which we mix the star ratings and qualitative to see if we can combine the qualitative and quantitative ratings to select the best funds. Control variables are also added to see the explanatory effect in relation to ratings. In the last empirical study, an innovative study is carried out, where the debate on the outperform, under-perform and neutral effect of sustainable investments is taken into account. Although there are some studies on this effect, no one has used these new ratings from Morningstar (they are fairly recent) and studied their effect on future performance of mutual funds. On the other hand, existing studies on known sustainable investments use only one dichotomous variable, which makes it impossible to perceive in what level, sustainable investments are beneficial, neutral or harmful to create value for mutual funds. 3. Structure of the study This thesis is divided into five chapters. Chapter I consists in a description of what it is proposed in this study: the reason behind our interest, it’s contribution to the knowledge of this topic, the goals to achieve, making a description the hypothesis and the methodology used. Chapter II is the first empirical study that is based on the predictive power of Morningstar’s star ratings divided into: introduction, literature review, a background of ratings and performance measures, methodology adopted, empirical test and conclusions. Chapter III is an empirical study about investments predictive power of qualitative ratings (analyst ratings). A study is made about the predictive power of Morningstar’s analyst ratings in the mutual fund’s performance. The structure is as follows: introduction; literature review; a background of ratings and performance measures; methodology adopted; empirical test and conclusions. Chapter IV it is another empirical study related to the sustainability scores of the portfolio of a mutual fund and its performance. We study the effect of investments socially responsible using the Morningstar’s Sustainability and ESG Scores followed by this structure: introduction; Are the ratings useful tools selecting mutual funds? 28 literature review; a background of ratings and performance measures; methodology adopted; empirical test and conclusions. The last one, Chapter V, is summary of the conclusions drawn from the three empirical studies, as well as all conclusions drawn and future research. 4. Objectives Many investors try to diversify their portfolios by investing through mutual funds, so their selection of funds that will integrate a portfolio is an important question. Since ratings are used by most of investors to select mutual funds, the main objective of this thesis is to analyze the usefulness of Stars ratings, Analyst ratings and Sustainability and ESG Scores of Morningstar in fund selection. The use of the Morningstar ratings is because it is the most important agency in the mutual fund industry. This broad objective can be broken down into different specific objectives: • To analyze if Morningstar’s quantitative ratings (Star ratings) can explain future performance and downside risk. • To investigate if Morningstar's Analyst ratings predict future performance. • To investigate if the combination of the qualitative and quantitative aspects of the Morningstar ratings can predict future performance. • To analyze if mutual funds that invest in SRI have superior, inferior or neutral performance compared to conventional investments. 5. Hypotheses Testing In Chapter II we test the following hypotheses: • H1: Mutual funds with better Star ratings will have better performance in risk-adjusted returns. • H2: Mutual funds with better Star ratings will have lower Value at risk (VaR). In Chapter III we test the following hypotheses: • H3: Mutual funds with better Analyst ratings will have better performance in riskadjusted returns. • H4: Mutual funds with better Analyst ratings will lower Value at risk (VaR). • H5: Mutual funds with better Analyst ratings and better Star ratings will have better performance in risk-adjusted returns. Chapter I: Introduction 29 • H6: Mutual funds with better Analyst ratings and better Star ratings will have lower Value at risk (VaR). In Chapter IV we test the following hypotheses: • H7: mutual funds with better ESG and Sustainability Score have neutral performance compared. • H8: mutual funds with better ESG and Sustainability Score have lower Value at risk (VaR). Are the ratings useful tools selecting mutual funds? 30 6. Methodology The thesis is divided into five chapters: Chapter II “Are quantitative ratings useful tools selecting mutual funds?”; Chapter III "Does Morningstar Analyst Ratings Matters for Mutual Funds?"and Chapter IV: "Does sustainability Score impact in performance?". Throughout this work, the statistical program used was the statistic program Stata 12 and the database used in all empirical studies was Morningstar Direct TM. In the second Chapter, we use the analysis of Stars Rating of Morningstar. The first method we use to examine the out-of-sample predictive performance is panel data regression model (Pooled and Random effects). The methodology based on panel data can control individual effects with advantages like the reduction of colinearity and efficiency, among others (Baltagi, 2010). We also use this methodology with 1 and 3 year lags, in order to evaluate the predictive ability of Morningstar ratings for medium and short-term performance. For each term, we use specific measures and data to avoid autocorrelation problems. First, we estimate the models using exclusively ratings to assess the effect of using only this variable for funds selection, later, we include other control variables, like expenses, size or manager experience. Instead of Panel data and based on Chen and Huang (2011), we also use the Quantile Regression to extend the regression model to conditional quantiles of the different performance metrics because it is more appropriate for a heterogeneous mutual fund universe, where strategies and objectives can vary. This model let us capture information about the coefficients at different quantiles of the dependent variable given the set of endogenous variables (star rating). In addition, the conditional quantile regression developed by Koenker and Bassett (1978) deals well with skewed distributions of fund performance. In particular, we adopt the bootstrapping method proposed by Efron (1979). In the third Chapter, we analyze if the Analyst rating can help to identify products that will outperform their peers in the following after the initial rating. At the same time, we also include separately stars ratings to compare both alternatives, taking into account that investors take their decisions based on backward looking data base. Finally, we use both ratings to evaluate if taking decisions combining good stars and analyst mutual funds, can help in the selection of out performers. Firstly, we estimate some models to evaluate the sample performance based exclusively on the analyst rating and their main pillars. If analyst rating is a forward-looking measure that reflects the expectations of analyst about future performance in the long run, we expect that higher ratings will obtain higher future performance. We calculate out Chapter I: Introduction 31 of sample risk adjusted returns to 12 and 36 months after the initiation rating and then we regress to the different metrics using Gold, Silver and Bronze indicators variables. For 12 months models, we estimate a panel data regression model (random effects) and for 36 we use ordinary least squares (OLS) regression because can we only have one period of three years after the initial grade. The methodology based on panel data can control for individual effects with advantages like the reduction of colinearity and efficiency. In addition, we also use the Quantile Regression to extend the regression model to conditional quantiles of the different performance metrics. In the fourth Chapter, we use OLS regression that we only have sustainability data on November 2016. We analysed the performance and the risk of the mutual funds with ESG and Sustainability Scores data of the last two years based on the work of Wimmer (2012), which shows that the portfolios of the funds regarding sustainability do not vary to that term (similar to the previous chapters). Chapter II: Are quantitative ratings useful tools selecting mutual funds? 33 Chapter II: Are quantitative ratings useful tools selecting mutual funds? 1. Introduction Mutual funds are one of todays’ leading saving products. An important question which investors have to face is the selection of funds to construct their portfolios. Ratings are used by most of the investors to select mutual funds; in fact, many investors make their choices based exclusively on these scores. Some studies such as Faff, Parwada and Poh (2007) and Del Guercio and Tkac (2008), support the evidence that investors make their decisions based on ratings. Their results showed that the variations in the flows of mutual funds are affected specially by the change in ratings. Despite the fact much research has been done in the field of mutual funds, in the particular case of ratings there are few studies and their results are not conclusive. One important question to answer is the relative to the ability of ratings to predict the future performance of mutual funds. In this sense, the main question we want to assess in this study is if Morningstar ratings are reliable tools selecting funds. Authors such as Blake and Morey (2000) and Morey (2002, 2005) among others, studied the predictive power of quantitative ratings obtaining weak results. However, some recent studies have found that ratings may lead to the choice of the best funds, and ratings can predict, in some extent, future performances. Authors such as Morey and Gotesman (2006), Müller and Weber (2014) and Meinhardt (2014) have demonstrated through their studies that ratings may be able to predict the future performance of mutual funds. In this paper, we evaluate the ability of quantitative ratings to anticipate the future behaviour of the fund’s performance. In addition, is important to check what is considered as a good rating and if there are performance differences between all categories or only between bad and good ones. We also evaluate the ability to anticipate short and long-term funds´ performance. In addition, we analyse if ratings also have differences in terms of downside risk through the analysis of VaR (value at risk). Finally, we include the interaction with costs and other variables such as size, management quality or categories, among others. Our results support the ability of quantitative ratings to predict future performance in the short and medium term even after the inclusion of management fees. Moreover, the best ratings have a better behaviour in terms of VaR helping to preserve the investors´ wealth. This work is important to the academic world, but also it is interesting for investors, rating companies and financial institutions. Are the ratings useful tools selecting mutual funds? 40 Table 2. Literature review of predictive power of ratings (*)$Methodological,paper. Authors Sample Number Predictive Power Conclusions Blake and Morey (2000) Two samples: Seasoned funds 1992-1997 and complete funds 1993 Morningstar On-Disk and Principia disks. U.S. domestic equity funds with Morningstar rating. 1993 sample group (635 funds). No - Is relatively easy to predict poor performance, but it is much more difficult to predict superior performance. - Low ratings indicate a poor performance. - “Naïve predictor” has more predictive power than 4 and 5 Stars. Morey (2002) September 1991 to September 2000 Morningstar On-disk or Principia. Morningstar’s Domestic Equity Category Depending on the year, ranging to less than 1,000 in 1991 to near 4,000 in 2000. No -Average overall star ratings of seasoned funds are consistently, and sometimes significantly, higher than the younger funds. - Methodology Morningstar is biased. - The weight and rounding used by Morningstar decline in overall ratings relatively more difficult for seasoned funds. - Funds ratings haven`t much ability to predict future performance Morey (2005) July 1993 to July 2001 All funds of Data Disk with rated a 5-star (273 funds) No - Expenses, Portfolio and Turn Over do not change much after received 5-Star Rating. -Three years after a fund received its initial 5-star rating, fund performance falls of. - Morningstar Ratings itself having a fantastic influence in fund flows. Morey and Gottesman (2006) July 2002 to June 2005 3,886 funds of Morningstar Principia mutual funds data of domestic equity funds Yes -The study tries to understand how the new Morningstar rating System, predicts future performance - Authors find that new system Morningstar Ratings can predict future performance in three years out the sample - Higher rated funds significantly outperform lower rated funds - The next to lowest rated funds (2star funds) significantly outperform the lowest rated funds (1Star funds) Del Guercio and Tkac (2008) November 1996 to October 1999 Morningstar s Principia 3,388 domestics equity mutual funds – Morningstar, Inc. -Authors argue that only 5-Star ratings can attract new assets - Excluding the rating initiation categories, flows change with downgrade and upgrade of Star Ratings Duret et al. (2008) (*) (*) No - Using a Markov modelling authors show that ratings persistence is poor Antypas et al. (2009) January 1998 to “paper date” Morningstar Direct Data 1,511 Equity funds quoted in US Dollars Yes - Better performances in Star3, Star 4, Star5 reflects stock selection rather than market timing abilities. - It´s more common that the worst funds with Star1, Star2 will also be worst in the future. So, Morningstar ranking system is most effective in identifying the worst-performing funds rather than the best-performing ones. Füss et al. (2010) May 2004 to April 2009 2,490 funds of Morningstar Inc. No - Results suggest that Morningstar Ratings have a little ability predicting future performance. - It`s possible to identify funds that will have poor performance in the future. - One-star ratings have worse results than the five-star category. - Morningstar rating has a big correlation with three-year Sharpe ratio. - Sharpe-ratio can be better at predicting future performance than Morningstar ratings Müller and Weber (2014) December 2001 to June 2008 Stiftung Warentest fund rating system (2,351 mutual funds for 30.06.2008) Yes - Past performance is positive related with future performance, in many funds. - This study reveals significant differences in persistence between different funds categories. - Predictive power depends on active or passive strategies. Chotivrtthamrong (2015) 2003-2007 36 Thai mutual funds form Morningstar Yes - Paper suggests a positive relationship between Morningstar Rating and performance. Chapter II: Are quantitative ratings useful tools selecting mutual funds? 41 4. Empirical study Our sample contains 1,579 European equity funds rated by Morningstar from 2004 to 2014. We limit our sample to the funds included in the following Morningstar categories: Europe Flex-Cap Equity, Europe Large-Cap Blend Equity, Europe LargeCap Growth Equity, Europe Large-Cap Value Equity, Europe Mid-Cap Equity, Europe Small-Cap Equity, Europe ex-UK Large-Cap Equity, Europe ex-UK Small/Mid-Cap Equity, Eurozone Flex-Cap Equity, Eurozone Large-Cap Equity, Eurozone Mid-Cap Equity and Eurozone Small-Cap Equity. The sample contains 10,375 observations, where 6.89 % have 1 Star, 19.37% have 2 Stars, 35.89% with 3 Stars, 26.46 % with 4 Stars and finally 11.38 % with 5 Stars. The funds are the type "open funds" with quantitative overall rating Morningstar in the investment area of total Europe. Investment area identifies the geographic area that the fund focuses its investments in. Funds selection has been done for all funds independently they were active or not to avoid survival bias. Furthermore, to avoid problems of multicollinearity, we select only a class for each fund, choosing these sequence preferences: institutional class, lower management fee, the lower net expense ratio, higher class size, oldest start date and accumulation preferred. In Table 3 the distribution of the rating is showed. Table 3Funds in the sample classified by Stars (Rating overall) Rating Overall Freq. Percent 1 Star 760 7.33% 2 Stars 2,169 20.91% 3 Stars 3,749 36.13% 4 Stars 2,575 24.82% 5 Stars 1,122 10.81% Total 10,375 100% 4.1. Performance Metrics To measure out-of-sample performance we use two risk-adjusted metrics: Alpha and Sharpe ratio. In addition, we include a non-adjusted one (annual return of the mutual fund) and the Value at Risk (99%) as a measure of downside risk. We will now explain briefly the -performance metrics: a) Sharpe Ratio is a way of measuring the expected return per unit of risk for a zero-investment strategy. Since it was introduced by William Sharpe in the 1960s, the Sharpe ratio has become one of the most widely used metrics in finance and economics. The Sharpe ratio is one of the tools most widely used metrics in finance and economics that help investors evaluate the relationship Are the ratings useful tools selecting mutual funds? 42 between risk and return of asset. By quantifying both volatility and performance, this tool allows for an incremental understanding of the use of risk to generate return. The Sharpe ratio can be represented by: 𝑆!" =(𝑅!" ! 𝑅! ) 𝛿 Where: • 𝑆𝑖𝑡 is adjusted-risk return • R 𝑖𝑡 represent the fund return • 𝑅 𝑓 represent the risk free • 𝛅 is the standard deviation b) We can consider Single-index alpha a return measure risk-adjusted, widely used to measure the returns of mutual funds. In fact, Alpha measures the return investment relative to the benchmark in which the assets are located. That means that single-index model is a simple way to measure a return of a portfolio, that it is represented by: 𝑅!"!𝑅!= 𝛼!+𝛽! (𝑅!" − 𝑅!)+𝜀!" Where: • 𝑅!"!𝑅! is the excess return on the market • αi is the abnormal return • βi is the stocks’ beta • 𝜀I are the residual (random) returns c) Annual Return is a metric to measure returns of Portfolio where the value is comparing with the time horizon previous. AR= !!"! !!"!! !!" Where: • Rit=Value of Portfolio • Rit-1= Value of Portfolio in t-1 Table 4 contains one year performance ahead conditioned to the previous rating. We can observe that generally highest-rated funds outperform persistently the worst rating in Chapter II: Are quantitative ratings useful tools selecting mutual funds? 43 terms of Sharpe, and in most of cases in terms of yearly return and Alpha, with the exception of 2008 and 2009. Table 4Conditional 1 year performance by Rating and Year This table reports the values of the dependent variables considered in the analysis obtained from Morningstar direct database. Sharpe Ratio is calculated in an annual basis, Alpha is the beta-adjusted return over one-year period, Return is the net yearly return. 4.2. Models The first method we use to examine the out-of-sample predictive performance is panel data regression model (Pooled and Random effects). The methodology based on panel data can control individual effects with advantages like the reduction of collinearity and efficiency, among others (Baltagi, 2010). In this sense, we estimate the following equations: 𝑌 !" =𝛼!+𝛽!𝐷4!" +𝛽!𝐷3!" +𝛽!𝐷2!" +𝛽!𝐷1!" +𝑌𝑒𝑎𝑟 ! ! +𝐶𝑎𝑡𝑒𝑔𝑜𝑟𝑦! ! +𝜀!" where: 𝑌 != performance metric for fund 𝑖. 𝑖= 1 through N, where N is the total number of funds in the sample. D4 = 1 if the fund received a 4-star overall Morningstar rating, 0 if not. D3 = 1 if the fund received a 3-star overall Morningstar rating, 0 if not. D2 = 1 if the fund received a 2-star overall Morningstar rating, 0 if not. Stars 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Alpha 1"star" -2.73 -2.21 -0.80 -0.32 30.18 30.56 10.40 -5.20 -1.34 -1.10 -1.58 2"star" -3.41 -1.00 -2.32 00.51 10.57 10.89 -0.14 -2.97 10.37 10.38 -3.55 3"star" -1.90 -1.47 -2.54 00.10 10.78 10.61 00.93 -2.41 10.09 10.47 -1.87 4"star" -0.03 00.01 -2.87 00.10 20.27 30.44 10.49 -1.77 10.99 20.94 -1.31 5"star" -0.59 -0.98 -1.24 -0.57 10.64 10.63 30.42 -2.48 40.36 40.38 -0.46 " Sharpe 1"star" -0.46 -1.04 10.00 10.17 0.84 -0.71 -0.54 0.13 -0.12 0.13 0.89 2"star" -0.39 -0.94 10.12 10.50 10.12 -0.63 -0.44 0.32 00.12 00.29 10.08 3"star" -0.23 -0.88 10.24 10.73 10.13 -0.58 -0.38 0.39 0.25 00.44 10.32 4"star" -0.03 -0.80 10.52 10.85 10.19 -0.52 -0.30 0.49 00.38 00.56 10.46 5"star" 0.29 -0.74 10.86 20.10 10.26 -0.43 -0.20 0.65 0.60 0.71 10.73 " Return 1 year 1"star" 90.88 26.37 15.29 00.78 -42.67 38.41 90.87 -18.50 16.97 18.14 30.28 2"star" 11.67 27.78 17.84 30.64 -44.82 35.83 12.17 -15.02 18.70 18.25 20.65 3"star" 12.58 28.26 21.40 20.60 -45.07 34.67 12.42 -14.62 19.09 19.56 40.15 4"star" 14.58 30.01 22.58 1.27 -45.84 34.69 13.06 -14.42 20.00 19.22 40.38 5"star1" 17.04 30.59 24.94 0.56 -47.02 31.20 13.63 -12.06 20.50 19.37 50.76 Are the ratings useful tools selecting mutual funds? 44 D1 = 1 if the fund received 1-star overall Morningstar in the year, 0 if not. Year = dummy time variables Category= dummy Morningstar categories. α! and β!,β!,β!,and β! are parameters of the regression and 𝜀! the term error. We also use this methodology with 1 and 3 year lags, in order to evaluate the predictive ability of Morningstar ratings for medium and short-term performance. For each term, we use specific measures and data to avoid autocorrelation problems. First, we estimate the models using exclusively ratings to assess the effect of using only this variable for funds selection. Later, we include other control variables, like expenses, size or manager experience. Instead of Panel data and based on Chen and Huang (2011), we also use the quantile regression to extend the regression model to conditional quantiles of the different performance metrics because it is more appropriate for a heterogeneous mutual fund universe where strategies and objectives can vary. This model let us capture information about the coefficients at different quantiles of the dependent variable given the set of endogenous variables (star rating). In addition, the conditional quantile regression developed by Koenker and Bassett (1978) deals well with skewed distributions of fund performance. In particular, we adopt the bootstrapping method proposed by Efron (1979) and implemented in the software Stata 12. Given 𝑌 ! as the different performance metrics used in this paper, and 𝑋! as a vector of exogenous variables representing the rating of the fund, the quantile model can be written as: 𝑦!= 𝑋! ´𝛽!+𝑢!" Assuming that: 𝑄𝑢𝑎𝑛𝑡!(𝑦!|𝑋!)= 𝑋! ´𝛽! 𝑄𝑢𝑎𝑛𝑡!(𝑢!"|𝑋!)=0 Chapter II: Are quantitative ratings useful tools selecting mutual funds? 45 4.3. Results for rating models The estimation of panel data models presented in Table 5 shows the results of regressing the risk adjusted return with the analyst indicators variables after twelve months since the rating is available, through a robust random effects panel data model. As in previous works, the models were initially estimated without control variables to evaluate the effect of selecting funds based exclusively on the rating analyst. Subsequently, other explanatory variables such as costs, size and management experience have been included. In all cases, category and time control variables have been included. As you can see, the results depend on the metric used, but in general, only funds classified as gold show better 12-month performance than the "Not recommended" funds in terms of Sharpe's ratio. Over 36 months the results are more disappointing and the only significant sign is negative. Therefore, our analysis reveals the inability of analyst ratings to identify funds that outperform their peers except in the case of gold funds, where the results show a better performance than those classified as not recommended. The differences with previous studies may be due to the fact of considering in our study a different sample focused exclusively on US, the panel data methodology and the different period considered. Table 5 shows that most of rating lagged variables are very significant over the different performance measures within one year. In general, funds that have a better rating one year ago, on average over-performed the subsequent lower grade in the next period. The level of fit observed in most of the estimated models indicates that the rating in addition to have predictive power explains an important part of the fund´s performance. In addition, this predictive power is also noted for yearly total return, indicating that those funds with better risk-return also get superior absolute returns. Our results support, in general terms, the performance persistence of quantitative ratings in the short-term (1 year) indicating its validity as a criterion for selecting funds. On average 1-star ratings obtain -1.02% Alpha than those rated as 5-stars. This difference is also true for the rest of the ratings with values that range from -0.24% for 4 stars to 0.87% for 2 stars. The above is also true for the Sharpe ratio where the profitability of higher-ratings funds outperforms those of lower ones throughout the period analysed. Again, the predictive ability is maintained in the case of Annual Return. However, the differences are not as clear when they are compared with the best funds (4stars), resulting only significant for the Sharpe ratio. In this regard, funds with better ratings (4 Are the ratings useful tools selecting mutual funds? 46 or 5 stars) seem to have a better ex-post behaviour, being especially noticeable for those with 1 and 2 stars. Our findings are in line with previous research made by Ferson and Schadt (1996), Morey and Gottesman (2006), Antypas et al. (2009), Müller and Weber (2014) and Meinhardt (2014); where they obtain some evidence in behalf of performance persistence. Table 5Panel Data with 1 lag (rating overall) Pooled Random Variable Alpha Sharpe Return Alpha Sharpe Return l1.stars4 -0.3055 -0.1396*** -0.1975 -0.2425 -0.1295*** -0.2775 L1.stars3 -0.8635*** -0.2531*** -0.6172*** -0.7579** -0.2383*** -0.7796*** L1.stars2 -1.0345*** -0.3458*** -0.7152*** -0.8799*** -0.3250*** -0.9747*** L1.stars1 -1.2201*** -0.4799*** -1.6881*** -1.0219* -0.4493*** -2.0720*** yr2004c 1.1685*** -1.7454*** 6.2224*** -1.1630*** -1.7417*** 6.2118*** yr2005c 0.7394* -0.1804*** 21.5354*** 0.7406*** -0.1775*** 21.5385*** yr2006c -1.2699*** 0.2447*** 14.8326*** -1.2653*** 0.2486*** 14.8226*** yr2007c 1.6178*** -0.2375*** -1.8108*** 1.6255*** -0.2329*** -1.8267*** yr2008c 5.2850*** -2.2675*** - 47.3888*** 5.2937*** -2.2637*** -47.4052*** yr2009c 3.1783*** -1.9710*** 26.5907*** 3.1712*** -1.9676*** 26.5781*** yr2010c 3.6807*** -1.7810*** 6.7910*** 3.6746*** -1.7783*** 6.7846*** yr2011c -0.2465 -1.0265*** -17.8362*** -0.2471 -1.0243*** -17.8429*** yr2012c 4.0070*** -1.1377*** 14.9007*** 4.0044*** -1.1367*** 14.9021*** yr2013c 3.9122*** -0.8914*** 17.4709*** 3.9048*** -0.8912*** 17.4746*** Largecapblend -1.2183*** -0.0662*** -1.4546*** -1.2444*** -0.0658*** -1.4521*** Largecapgrwth 0.7505 0.0749*** 0.1501 0.7384 0.0795** 0.1458 Largecapvalue -1.6101*** -0.1492*** -2.2181*** -1.6335*** -0.1536*** -2.2043*** Midcap -0.0815 0.1446*** 3.0937*** -0.0962 0.1428*** 3.1154*** Smallcap 0.3016 0.1772*** 4.4938*** 0.29 0.1767*** 4.5258*** LargecapexUK -0.0228 0.0118 0.5125 -0.0443 0.0125 0.507 SmallmidexUK 1.8133* 0.1417*** 3.1985*** 1.8002* 0.1483*** 3.2359** Eurozoneflexcap -2.6167*** -0.1790*** -2.2755*** -2.7362 -0.1818*** -2.2626*** Eurozonelargecap -1.1144*** -0.2035*** -2.7671*** -1.1366** -0.2088*** -2.7594*** Eurozonemidcap 2.0759* -0.0073 1.6261** 2.1472 -0.0081 1.6309** Eurozonesmallcap 1.7764* -0.0285 0.8476 1.7516 -0.03 0.8754 cons -0.4619 1.7133*** 5.6667*** -0.5264 1.7005*** 5.8201*** N 4435 8776 8860 4435 8776 8860 r2_w - - - 0.1771 0.9317 0.9184 Rho - - - 0.0867 0.2384 0.00 Chi-square (p) 0.000 0.000 0.000 - - - This table reports the coefficients for Panel Data pooled and random models for different performance measures. Alpha is the beta-adjusted return over a one-year period; Sharpe is the yearly risk-adjusted return and, Return is the total one year net return. L1. Star is the one year lagged variable representing the rating of the mutual fund and yr* are the time dummies variables and finally, Largecapblend, Largecapgrwth, Largecapvalue, Midcap, Smallcap, LargecapexUK, SmallmidexUK, Eurozoneflexcap, Eurozonelargecap, Eurozonemidcap and Eurozonesmallcap are dummies to control for categories. N is the number of observations, r2 the pseudo-squared fit measure, Rho is the fraction of variance due to individual effects and Chi-square (p) is the p-value associated to the Chi-square significance test. *Significant at 10%; ** significant at 5% and *** significant at 1%. Chapter II: Are quantitative ratings useful tools selecting mutual funds? 47 As can be seen in the table below (Table 6), the results in the 3 lags pooled and random effects models confirm previous achievements in the pooled model and partially in the random effects. However, the differences in performance reduce considerably and in many cases, they aren´t significant for the random effects model. The results support the significance in shorter time periods, and to a lesser extent, in longer periods. Something that seems logical since in the long term it is likely that the fund rating will change, and consequently, this can affect their future performance. Are the ratings useful tools selecting mutual funds? 48 Table 6 - Panel Data with 3 lags (rating overall) Pooled Random Variable Alpha Sharpe Return Alpha Sharpe Return L3.stars4 -0.1861 -0.0430** -0.0186 0.0006 -0.0314 0.1198 L3.stars3 -1.0433*** -0.0881*** -0.7414*** -0.7006* -0.0620*** -0.4324 L3.stars3 -1.2857*** -0.1000*** -0.7216** -0.7095 -0.0522** -0.1951 L3.stars1 -1.7105*** -0.1982*** -1.7070*** -1.0025 -0.1276*** -0.9342** yr2004c -0.8695*** 0.6627*** 7.7480*** 0 1.5507*** 21.9404*** yr2007c 0.2321 -0.8898*** -14.2031*** 1.0776*** 0 0 yr2010c - - - 0.8142*** 0.8861*** 14.1835*** Largecapblend -0.5737 -0.0094 -0.5852* -0.6119 -0.0118 -0.587 Largecapgrwth 1.1552* 0.1480*** 1.2841** 1.1384* 0.1424*** 1.2674* Largecapvalue -1.1131** -0.0752*** -1.3074*** -1.1195** -0.0832** -1.3268*** Midcap -0.0681 0.1084*** 2.7028*** -0.0483 0.1156*** 2.7358*** Smallcap 0.0362 0.1292*** 3.5137*** 0.0506 0.1309*** 3.4996*** LargecapexUK 0.2941 0.0911*** 1.4438*** 0.2126 0.0839*** 1.4289*** SmallmidexUK 1.8214* 0.1209** 3.1604*** 1.6888 0.1247* 3.1690*** Eurozoneflexcap -0.8713 -0.1648*** -2.2805*** -1.3277 -0.1764*** -2.3472** Eurozonelargecap -0.2264 -0.0873*** -1.5137*** -0.229 -0.0967*** -1.5212*** Eurozonemidcap 1.2723 0.0057 1.4497* 1.5739 -0.0026 1.3626* Eurozonesmallcap 1.7389 -0.0066 0.6263 1.6054 -0.0094 0.616 cons 0.8002* 0.5914*** 8.3063*** -0.3053 -0.3167*** -6.1738*** N 1510 2989 3017 1510 2989 3017 r2_w - - - 0.1321 0.9131 0.9077 Rho 0.00 0.00 0.00 - - - Chi-square (p) - - - 0.01 0.00 0.00 This table reports the coefficients for Panel Data pooled and random models for different performance measures. Alpha is the beta-adjusted return over a one-year period; Sharpe is the yearly risk-adjusted return and, Return is the total one year net return. L3.Star is the tree year lagged variable representing the rating of the mutual fund and yr* are the time dummies variables and, finally, Largecapblend, Largecapgrwth, Largecapvalue, Midcap, Smallcap, LargecapexUK, SmallmidexUK, Eurozoneflexcap, Eurozonelargecap, Eurozonemidcap and Eurozonesmallcap are dummies to control for categories. N is the number of observations, r2 the pseudo-squared fit measure, Rho is the fraction of variance due to individual effects and Chi-square (p) is the p-value associated to the Chi-square significance test. *Significant at 10%; ** significant at 5% and *** significant at 1%. Table 7 shows the results of quantile regression with 1 lag confirming previous achievements for the different percentiles. In particular, the positive difference for better ratings in risk-adjusted performance is confirmed for all the levels considered. However, the differences are more significant in the sample of the most profitable funds and those who are around average, than in the 75 percentile (Q75), when we take the total return. Chapter II: Are quantitative ratings useful tools selecting mutual funds? 49 Table 7Quantile regression for 1 and 3 lags (rating overall): Equation 1 Variable Alpha (Q25) Sharpe (Q25) Return (Q25) Alpha (Q50) Sharpe (Q50) Return (Q50) Alpha (Q75) Sharpe (Q75) Return (Q75) l1.stars4 -0.7856** -0.1031*** 0.2001 -0.6863*** -0.1494*** -0.4111 -1.2358*** -0.1994*** -0.7705** L1.stars3 -1.2799*** -0.1953*** -0.0843 -1.4531*** -0.2599*** -1.1650*** -2.2646*** -0.3041*** -1.6594*** L1.stars3 -1.4429*** -0.2989*** -0.5981 -1.5744*** -0.3502*** -1.4717*** -2.5332*** -0.3958*** -1.9874*** L1.stars1 -2.9872*** -0.4588*** -2.2046*** -2.4037*** -0.4694*** -2.4399*** -1.8628*** -0.4967*** -1.9123*** yr2004c -2.7910*** -1.5629*** -" -1.6996*** -1.5510*** -" -0.8632** -1.6013*** -" yr2005c -" -" 15.3620*** -" -" 15.3565*** -" -" 15.1855*** yr2006c -3.0676*** 0.4041*** 8.5395*** -1.8229*** 0.4826*** 8.6942*** -0.5770** 0.4987*** 8.6621*** yr2007c -0.3445 -0.0779*** -8.3565*** 0.4629 -0.0082 -7.3436*** 2.1939*** 0.0554** -5.1723*** yr2008c 2.1939*** -2.0514*** -53.7852*** 3.7600*** -2.0556*** -53.0733*** 6.4822*** -2.0776*** -52.2483*** yr2009c 0.9705*** -1.7414*** 17.9727*** 2.0200*** -1.7384*** 19.7295*** 3.8944*** -1.7856*** 22.2340*** yr2010c 1.1888*** -1.5382*** -2.8859*** 2.3813*** -1.5638*** 0.3644 3.7456*** -1.6391*** 2.7241*** yr2011c -1.4034*** -0.8576*** -24.1165*** -0.6318** -0.8110*** -23.1419*** 0.4001* -0.8188*** -22.1111*** yr2012c 2.6613*** -0.9402*** 8.8923*** 2.6910*** -0.9146*** 9.0832*** 3.6628*** -0.9592*** 9.3028*** yr2013c 2.3230*** -0.6577*** 10.8681*** 2.7629*** -0.6576*** 11.3616*** 4.2652*** -0.7141*** 12.1108*** yr2014c -0.7831*** 0.1710*** -5.1934*** -0.4341* 0.2197*** -5.3938*** -0.1749 0.2416*** -5.7037*** Largecapblend 0.8840* -0.0154 0.8613*** -0.717 -0.0668*** -0.9523*** -2.6096*** -0.1168*** -2.6583*** Largecapgrwth 1.6389*** 0.0156 0.6766* 0.7471 0.0603** 0.3161 0.4759 0.0791* 0.0175 Largecapvalue 0.1818 -0.0775*** -0.651 -1.1429** -0.1275*** -1.4526*** -2.9139*** -0.1699*** -2.7086*** Midcap 0.6077 0.1284*** 1.4710** 0.0522 0.1691*** 3.9220*** -0.845 0.1449*** 4.9620*** Smallcap 0.7962 0.1692*** 1.2280** 0.5834 0.1823*** 4.7041*** 0.0315 0.2024*** 7.8527*** LargecapexUK 1.8062*** 0.0791*** 3.2123*** 0.456 0.0027 1.1272*** -1.1739 -0.0666*** -0.7071* SmallmidexUK 2.8893*** 0.1229*** 1.6649* 2.0117 0.0968 0.4185 2.3599 0.1651*** 5.2740** Eurozoneflexcap -1.8575 -0.1502*** -2.1267*** -1.4704 -0.1393*** -1.8200*** 0.9037 -0.1591*** -1.5396** Eurozonelargecap 0.8617 -0.1405*** -0.4805 -0.6594 -0.1855*** -1.8067*** -2.3432*** -0.2132*** -3.5288*** Eurozonemidcap 0.7202 -0.007 0.8945 4.5564*** 0.0098 1.1887 2.6893* 0.0236 2.7474** Eurozonesmallcap 2.1983 0.0015 -0.1535 1.8732* 0.0021 0.4545 0.6663 -0.0251 2.3513** _cons -2.6679*** 1.2847*** 6.9382*** 0.464 1.4816*** 11.3829*** 4.1102*** 1.7155*** 15.3961*** This table reports the coefficients for Quantile regression. Alpha is the beta-adjusted return over a one-year period; Sharpe is the yearly risk-adjusted return and, Return is the total one year net return. L1.Star is the one year lagged variable representing the rating of the mutual fund and yr* are the time dummies variables and, finally, Largecapblend, Largecapgrwth, Largecapvalue, Midcap, Smallcap, LargecapexUK, SmallmidexUK, Eurozoneflexcap, Eurozonelargecap, Eurozonemidcap and Eurozonesmallcap are dummies to control for categories. *Significant at 10%; ** significant at 5% and *** significant at 1%. Are the ratings useful tools selecting mutual funds? 56 significant in explaining performance, indicating that costs are not the only factor that determines the predictive power of quantitative ratings. Finally, the best ratings perform better in terms of VaR showing that the investment in good rated funds can help to preserve the investors´ wealth better. Our results support the use of ratings in the investment funds selection process, accompanied by other quantitative variables. On the other hand, the greater significance achieved in the short term advises the review of portfolios on an annual basis. Finally, the inclusion of qualitative factors can help improve the selection process of investment funds. Chapter III: Does Morningstar Analyst Rating matters for mutual funds? 57 Chapter III: Does Morningstar Analyst Rating matters for mutual funds? 1. Introduction Mutual funds grew explosively in recent times and the researches on this topic have multiplied. In fact, companies, banks, investors, and others are caring for mutual funds so it is a matter of interest that goes beyond the academic world. To make life easier for investors, rating agencies such as Morningstar, Lipper Leaders, MSCI, Standard and Poor's, among others, attached notes to mutual funds to help investors in selecting their funds. In fact, there are investors who make their decisions based exclusively on ratings. This is why many funds use their rating as advertising to attract investors. Most popular ratings are Morningstar Star ratings and some empirical studies as Blake and Morey (2000) and Guercio and Tkac (2008) and among others, have shown that the downgrade or upgrade of quantitative ratings have an influence in the flows of mutual funds. In addition, in the previous chapter we have shown the ability of Stars Rating (backwards looking) to explain the out-of-sample performance in the short term and to preserve the wealth of investors. Authors such as Morey and Gotesman (2006), Müller and Weber (2014) and Meinhardt (2014) have also supported the performance persistence of quantitative ratings but only in the short term. However, the selection of funds based exclusively on historical performance or quantitative ratings, excludes a set of qualitative factors that can explain future performance. This is why, in addition to quantitative ratings it have appeared ratings based on analyst opinions that evaluate features of mutual funds as: Governance, Process, People, Parent, Board Quality, Corporate Culture, Fees, Manager Incentives or Regulatory Issues. Despite there exists several studies about quantitative ratings, qualitative ratings are not as popular and very few research has been done about the ability to select good funds based on qualitative ratings. In the particular case of Morningstar, there are two alternatives: Analyst Ratings and Stewardship Grade. As far as we know, very few researches have been conducted on this subject. Wellman and Zhou (2008), Ng (2009), Lai, Tiwari and Zhang (2010), Chen and Huang (2011), Gottesman and Morey (2012) and Cao, Ghosh, Goh and Ng (2012) studied the effects on performance and flows of Stewardship Grades. On the other hand, Kamal (2013) and Are the ratings useful tools selecting mutual funds? 58 Armstrong, Genc and Verbeek (2016) are the unique authors that focus their research on Analyst ratings. Analyst rating was launched in September 2011 by Morningstar and it is a forwardlooking measure based on analyst’s expectation about the future performance of the mutual fund relative to the peers and for the long term. The rating reflects the valuation of analyst in five dimensions which includes factor like the cost, past performance, quality of management, interest alignment, etc. Armstrong et al. (2016) find “higher abnormal flows to funds receiving higher ratings suggesting that the average retail investor values the analyst’s subjective views when allocating their wealth”. Thus, investors take into account both stars (backward-looking) and analyst (forward-looking) to take their decisions of investment. In this paper we assess to what extent selecting mutual funds based on ratings criteria has an impact on the financial and risk performance of investors. In particular, we endeavour to answer three essential questions: (1) Do good analyst (forwardlooking) ratings outperform non-recommended ones in the short and long term? (2) Do good stars (backward-looking) ratings outperform bad ones in the short and long term? (3) Is useful to combine both ratings in the screening process to identify good future performers? Previous research on analyst ratings is scarce and, in general, includes all the universe of mutual funds with rating; despite the heterogeneity in terms of investment area, exchange rate risk, period of analysis, etc. This is why our research only contains the analyst ratings for rated funds from August 2012 to August 2016; because previous studies have been done with very limited samples or assuming questionable hypothesis about rating persistence. Further, the results of the literature about the ability of “good analyst rated” to outperform “bad analyst rated” have shown mixed evidence. The empirical research into these questions is of particular interest to asset managers, financial advisors and investors using ratings to take their portfolios decisions. Our results are in line with Kamal (2013) and Armstrong et al. (2016). In this way, we have found a small evidence that, on average, funds with a better Analyst Rating (Gold) have a better performance in terms of risk adjusted measures (alpha and Sharpe). The predictability is observed in several analyses done in one year ahead but not for three-years. This evidence is more relevant in the case of the analysis made by investment style´s category. In the analysis of the pillars in which the analyst ratings are Chapter III: Does Morningstar Analyst Rating matters for mutual funds? 59 broken down we do not find evidence that future performance is related to any of these specific dimensions. The paper is organized as follows. First, we describe Morningstar Analyst and Stewardship Grade methodology. Then, we review and summarise the main existing research about qualitative ratings and mutual fund’s performance. In the fourth section, we present the empirical analysis, the statistical models and the main results. Finally, we summarize the main conclusions. 2. Morningstar qualitative ratings. Morningstar has two systems to classify mutual funds based on qualitative aspects: Morningstar Analyst Rating and Morningstar Stewardship Grade. Stewardship Grade is determined using some quantitative measures, but it is primarily based on qualitative information across five areas: Corporate Culture, Fund Manager Incentives, Fees, Fund Board Quality and Regulatory History. The Morningstar Stewardship Grade for Funds assigns a letter grade from A (best) to F (worst) for each fund. Morningstar Analyst Ratings are forward-looking qualitative and quantitative analyses of mutual fund about five pillars: Process, Performance, People, Parent and Price. Analyst Ratings are based on the convictions that funds will outperform their benchmarks over the long term. 2.1. Stewardship Grade In 2003, a series of scandals related to the management (late trading, market timing and other irregularities) affected US mutual funds (see for example Bogle, 2010 for a summary). The importance of this issue took a series of regulatory reforms, but also rating agencies began to focus more on the issue. In August 2004, Morningstar launched Fiduciary Grades for funds, in 2005 renamed Stewardship Grades. These grades provide a standard of corporate governance ranging from A (best) to F (worst). Stewardship Grades are calculated as the aggregate scores of five components – Corporate Culture, Board Quality, Manager Incentives, Fees and Regulatory History-. Morningstar (2010) reveal the details of the methodology for the Mutual Fund Stewardship Grade. One important changes in the methodology is that in 2011, Morningstar began to use Stewardship Grades in Analysts Ratings because they have better information about Parent pillar. Another important change is that Stewardship Grade changed the weight of the components corporate culture and fund manager Are the ratings useful tools selecting mutual funds? 60 incentive because they found more predictive power in an empirical study. The new grade now applies to fund companies rather than to individual funds only. The scoring of the Stewardship Grades makes a scale that`s are graded from A (best funds) to F (Worst fund´s). The grades depending all the time of their specified relative cultures. Funds that have a superior culture will standard; but that haven´t all the best practices receive B. Fund companies that meet industry standards receive D or F grades. The maximum global score is 10 points, and is based on the sum of the five component scores: A (9–10 points). B (7–8.5 points), C (5–6.5 points), D (3–4.5 points) and F (2.5 points or fewer). The Corporate Culture component’s maximum score is 4 points. For the Board Quality, Manager Incentives, and Fees components, the maximum score is 2 points, and points are awarded in increments as 0.5 points. For the Regulatory History component, the maximum score is 0 points, and the lowest possible score is -2 points. Regulatory History scores may be reduced in increments as 0.5 points. The components that Morningstar evaluate are the follow: 1) Regulatory Issues: examine any regulatory issues fund with time horizon at the last past 3 years. 2) Board Quality: focus on the quality of fund`s board, looking at multiplies factors. 3) Manager Incentives: evaluate two aspects. The first is the fund ownership, when they pretend understand if the manager has significant investment in the fund, if the funds run by the manager are inappropriate for such a large investment, among others. Second is the compensation structure, if exist incentive programs that encourage a focus a short-term performance or asset growth are viewed less favourably. 4) Fees: this component pretends to understand two aspects. First, If the fund`s expense ratio is below the average for its type of share class and second if the fund`s expense ratio declined meaningfully as assets have grown. 5) Corporate Culture; this component looks at how seriously a firm takes its fiduciary duty to its fund´s shareholders. Chapter III: Does Morningstar Analyst Rating matters for mutual funds? 61 2.2. Morningstar Analyst Rating Morningstar Analyst Rating, created by Morningstar in November 2011, are forward-looking qualitative and quantitative analyses of mutual funds. The Analyst Rating is expressed as metals: Gold, Silver, Bronze, Neutral and Negative. The medals Gold, Silver, Bronze are the notes for better funds (Recommended) and the Neutral and Negative are notes by worst funds (Not Recommended). Analysts evaluate funds based on five key pillars: Process, Performance, People, Parent and Price. These keys pillars are the components that Morningstar analysts believe that may be the predictors of outperform over long-term on a risk-adjusted basis. Thus, Morningstar Medallists— Gold, Silver, or Bronze rating—are funds that Morningstar analysts believe will perform better over time compared to similar investments (category group) in the long run (five years). Morningstar (2011) summaries the methodology: 1) Process: analysts try to understand the strategy and how management has competitive advantage to run the process well and consistently. 2) Performance: analysts try to understand what is fund`s strategy and the pattern logical given its process. Another thing is understanding if a fund has strong risk-adjusted returns over a relevant time. 3) People; understand manager´s talent, tenure and resources. 4) Parent; understand if firm prevail Salesmanship or Stewardship (e.g. create alignment of interests with their clients, good governance or have a long-term investment horizon, charge reasonable fees, etc.). 5) Price; due that Morningstar knows that expenses are one of the better predictors of future outperformance this pillar try to resume if a fund have reduced cost compared with similar funds sold through similar channels. Each pillar is rated positive, neutral or negative. Morningstar Analyst Ratings are based on overall analysis and ratings of the five pillars. Morningstar (2011) describes Analyst Ratings in the follow order: 1) Gold; fund distinguished between the five pillars and have guarantee of highest level of conviction that will perform better over time compared to similar investments. 2) Silver, fund with sufficient level of conviction to guarantee a positive rating and the advantages are bigger than disadvantages in the five pillars. 3) Bronze; fund witch notable advantages across several, but not all of pillars. Are the ratings useful tools selecting mutual funds? 62 4) Neutral; funds that don’t have a strong positive or negative conviction. but they aren’t likely to seriously underperform their relevant performance. 5) Negative; fund that Analyst Rating consider an inferior offering to its peers and that has reason likely to significantly underperform (e.g. high fees or an unstable management team). Morningstar may also use two other designations in place of a rating: Under Review and Not Rateable. Under Review means that the fund requires further review to determine the impact on the rating. Not Rateable is used in the case there are no relevant comparators. Haslem (2014) summaries several works published in the Morningstar Fund Investor, a monthly newsletter of Morningstar, that explain several aspects of the construction and implications of Analyst Ratings and Stewardship Grades (among others). At the same time, the author summarizes the main research assessments of Morningstar ratings on mutual fund performance. 3. Previous research. There is a debate in literature about the power of ratings to predict future performance. Most of the studies employ quantitative ratings in order to look at the capacity of the ratings by choosing the best funds. The quantitative aspects of ratings have some limitations. Quantitative ratings cannot quantify aspects of qualitative ratings. It is true that when you build ratings based on historical performance in some way implied the ability of managers, as well as other qualitative aspects. However, there is no evidence that past performance has this ability to capture the ability of managers to get results above the market. On the other hand, there are aspects that have to do with ethics, correct procedures, appropriate strategies, as well as whether fund managers are hedging the fund's wealth in the medium and long term. Certain qualitative aspects such as quality of management and procedures cannot be captured by past performance and it is in this sense that qualitative ratings can give very useful information to choose the best funds. Tufano and Sevick (1997) is the first empirical study examining quality of the board of directors at mutual funds. They examine the relationship between the composition and compensation of boards of directors of U.S. mutual funds and the fees charged to their investors. In the last years, authors as, Del Guercio, Dann and Partch (2003), Qian (2006), Meschke (2007), Khorana, Servaes and Wedge (2007), Khorana, Chapter III: Does Morningstar Analyst Rating matters for mutual funds? 63 Tufano and Wedge (2007), Ferris and Xuemin (2007), Boyd and Yilmaz (2007), Trahan (2008), Evans (2008), Kong and Tang (2008), Cremers, Driessen, Maenhout and Weinbaum (2009), Adams, Mansi and Nishikawa (2010), Chou, Ng and Wang (2011), Ding and Wermers (2012), Hazenberg (2012), Kryzanowski and Mohebshahedin (2016) or Mamatzakis and Xu (2017) investigated the quality of governance at mutual funds and they found that in general, funds with better governance obtain better future performance. There are other papers that focus on the influence of Stewardship Grade or Analyst Rating on fund flows. Wellman and Zhou (2008) found that investors sell funds with poor Stewardship Grades and buy those with good grades. Lai, Tiwari and Zhang (2010) suggest that investors react more strongly to poor fund performance by withdrawing funds when the board quality component of Morningstar’s Stewardship Grade is perceived to be bad. Armstrong et al. (2016) concluded that higher Analyst ratings (Gold and Silver) receive higher abnormal flows1, particularly in retail funds. These studies report evidence on the importance for qualitative ratings to investors. The relative novelty of qualitative ratings of Morningstar causes that, unlike investigation of Star Ratings, there is not too much research on this topic in the literature and just few authors have studied the subject. In the case of Stewardship, we highlight Gerrans (2006), Wellman and Zhou (2008), Ng (2009), Lai, Tiwari and Zhang (2010), Chen and Huang (2011), Gottesman and Morey (2012), Cao et al. (2012). For Analyst, only Kamal (2013) and Armstrong et al. (2016) had studied their effect on future performance. Gerrans (2006) investigated the relationship between Morningstar Star ratings, a qualitative rating (QL) and their product in the performance of Australian managed funds. QL rating was an assessment of fund administration, investment management, product and company capabilities and strengths reported as Business and Management Strength rating and Sector Strength rating. He employed data form two of the largest fund subcategories Australian Equity Trusts–General (AET) and Superannuation– Australian Equity Trusts (SAET). Data were collected from the Morningstar Total Access CD between August 1996 and February 2001. The results do not provide evidence to support a positive relationship between ratings and four commonly used 1 Abnormal flows are computed as the difference between the flow to the rated fund and the flow to a ”matched” unrated fund within the same style classification using propensity score matching (PSM). Are the ratings useful tools selecting mutual funds? 64 performance measures (geometric monthly return, onefactor alpha, four-factor alpha and Sharpe ratio). Wellman and Zhou (2008) is the first work in study future performance employing Stewardships. They concluded that there are significant differences in performance between mutual funds that have good Stewardship Grades and those who have bad. Funds with top Stewardship Grade (“A” or “B”) outperform those with poor grades (“D” or “F”) by 19 to 23 basis points per month over the period analysed January 2001 - July 2004, and by 10 to 16 basis points over the period September 2004 – December 2004. In the 27 months after the announcement of the grades, good funds outperformed bad funds by a significant 10 basis points. They find that among the five stewardship components, only Fees and Board Quality exhibit significant explanatory power. Regulatory History, Manager Incentives, and Corporate Culture show no explanatory power in explaining ex-post risk-adjusted returns. Ng (2009) examines applying least-squared regressions and multinomial ordered logit regressions the extent to which Morningstar Star Ratings and Morningstar Stewardship Grades can predict future fund performance. In particular, he investigates the combined predictive power of the two ratings in a twelve-month sample (January 2005 - December 2005). His results show that none of the ratings alone possesses strong predictive power, but the combined rating is superior in forecasting future returns. Lai, Tiwari and Zhang (2010) focus on the board quality component of Morningstar’s Stewardship Grade. They find a significant relationship between board quality and performance persistence. For funds with board quality, negative past performance predicts future negative performance and also there is evidence of shortterm persistence in positive performance (positive past performance predicts future positive performance). Chen and Huang (2011) study the relationship between the performance and Stewardship grades of Morningstar using the methodology of the OLS and quantile regression from the 2006 to 2009 (2nd quarter). In the study, the authors used the overall Stewardship Grade and by another hand, two components grades (Manager Incentives and Board Quality). The statistical results indicate Stewardship Grades are strongly positively related to fund performance measured by the Sharpe Ratio. The OLS regression reveals a strong association between overall Stewardship Grade and the fund performance, but it is not so clear that all the components can have the same predictive power. Quantile regressions show that there is a strong relationship with the right tail of Chapter III: Does Morningstar Analyst Rating matters for mutual funds? 65 the performance distribution. Authors also show that Stewardship Grades are strongly negatively related to portfolio turnover. Finally, this study shows that there is little ability to predict future performance of Alfas. Gottesman and Morey (2012) have studied if Stewardship Grade can predict future performances. They tested the capacity of their components to predict riskadjusted performance of domestic equity funds over the period 2005-2010. The authors used methods that are robust to survivorship bias and they find that corporate culture have little predictive power on future performances. They also find that no one component of Stewardship Grade can predict fund performance consistently. Cao et al. (2012) test the role of Morningstar Stewardship Grade in mutual fund performance. They use data form Morningstar Direct and Centre for Research in Securities Prices (CRSP) Survivorship Bias Free Mutual Fund database over the sample period November 2004 – May 2011. Their findings suggest that corporate governance grades of mutual funds carry information for predicting long-term mutual fund performance. Research on Morningstar Analyst Rating is recent and limited. To our knowledge only Kamal (2013) and Armstrong et al. (2016) have analysed the effect on Morningstar Analyst on future performance. Kamal (2013) studied Morningstar Analysis Ratings at July 2013 (1,159 individual mutual funds: equity fund, fixed-income funds, etc.) and concluded that there is a significant positive relationship between these ratings and the future performance as measured by the 3-year Alpha applying OLS and quantile regression. Results for quantile regression show that for better performing funds; higher Analyst Rating does not necessarily predict better performance in the future. He also found that the People pillar of these ratings has a significant predictive power for funds’ future performance. However, the author cautions in her work that “whether the Analyst Ratings can predict future fund performance, we need more data, which is not available as of yet, because these are relatively newer ratings, with a long-term focus. Finally, she found that Sharpe Ratio and Analyst Ratings are significantly positively related to contemporaneous fund performance. The People and Process Ratings are also individually significantly related to the Sharpe Ratio. Armstrong et al. (2016) tried to understand to what extent if Morningstar Analysis Ratings have the power to influence investors in terms of flows, as well as these ratings have the ability to provide above average performances. The authors Are the ratings useful tools selecting mutual funds? 72 data can control for individual effects with advantages like the reduction of collinearity and efficiency, among others (Baltagi, 2010). The following equations are estimated: 𝑌 !,!!!=𝛼!+𝛽!𝐺𝑜𝑙𝑑!" +𝛽!𝑆𝑖𝑙𝑣𝑒𝑟 !" +𝛽!𝐵𝑟𝑜𝑛𝑧𝑒!" +Month! ! +𝐶𝑎𝑡𝑒𝑔𝑜𝑟𝑦! ! +𝜀!" 𝑌 !,!!!=𝛼!+𝛽!𝐺𝑜𝑙𝑑!" +𝛽!𝑆𝑖𝑙𝑣𝑒𝑟 !" +𝛽!𝐵𝑟𝑜𝑛𝑧𝑒!" +𝛽!𝑁𝑒𝑡𝑒𝑥𝑝!" +𝛽!𝑙𝑜𝑔𝑆𝑖𝑧𝑒!" +𝛽!𝑀𝑎𝑛𝑎𝑔𝑒𝑟𝑇𝑒𝑛!" +Month! ! +𝐶𝑎𝑡𝑒𝑔𝑜𝑟𝑦! ! +𝜀!" Where Y!,!!! is the performance obtained by the fund i for 12 of 36 months after the initial rating. As performance measures we use Alpha, Sharpe and Total Return. Gold, Silver and Bronze are indicator variables considered as recommended funds, that take the value of 1 when the mutual fund is rated as one of these categories and 0, otherwise. Category are dummy variables for the nine categories considered in the study. Finally, 𝛼! and 𝛽!,𝛽!,𝛽!, 𝛽!, 𝛽!, and 𝛽! are parameters of the regression and ε! the term error. We estimate the models first controlling by month and categories, and then including some additional controls like expenses, size and age. Following Chen and Huang (2011) and Armstrong et al. (2016), we also include as control variables the manager tenure (ManagerTen) of the fund, the costs measured by the net expenses ratio (Netexp) and the size of the mutual fund (LogSize). 4.3. Results for Analyst ratings (forward looking) Table 17 shows the results of regressing the risk adjusted return with the analyst indicators variables after twelve months since the rating is available, through a robust random effects panel data model. As in previous works, the models were initially estimated without control variables to evaluate the effect of selecting funds based exclusively on the rating analyst. Subsequently, other explanatory variables such as costs, size and management experience have been included. In all cases, category and time control variables have been included. As you can see, the results depend on the metric used, but in general, only funds classified as gold show better 12-month performance than the "Not recommended" funds in terms of Sharpe's ratio. Over 36 months the results are more disappointing and the only significant sign is negative. Therefore, our analysis reveals the inability of analyst ratings to identify funds that outperform their peers except in the case of gold funds, where the results show a better performance than those classified as not recommended. The differences with previous studies may be due to the fact of considering in our study a different sample focused exclusively on US, the panel data methodology and the different period considered. Chapter III: Does Morningstar Analyst Rating matters for mutual funds? 73 Table 17Analyst ratings and out-of-sample performance after 12 and 36 months Variable Alpha-12 Sharpe-12 AlphaC-12 SharpeC-12 Alpha-36 Sharpe-36 AlphaC-36 SharpeC-36 Gold 0.7204 0.2023*** 0.7288 0.3572*** 0.311 0.0368 0.024 -0.012 Silver -0.1788 0.0219 -0.2108 0.1581 -0.3122 -0.0945* -0.1202 -0.0711 Bronze -0.1094 -0.0334 0.4072 0.1216 -0.1893 -0.047 0.1439 -0.0055 Netexp_ - - -0.7281 -0.0824 - - -0.8025 -0.1673*** logSize - - 0.2006 -0.0123 - - 0.4997*** 0.0520*** ManagerTen_ - - -0.0118*** -0.0039*** - - -0.0095** -0.0009** USlargeblend -1.8096*** 0.2660*** -2.7008*** 0.2364*** -2.2694*** 0.1259* -2.8618*** 0.0785 USlargegrowth -2.6385*** 0.1472** -2.0243*** 0.2421*** -0.9392 0.2351*** -1.3944** 0.1900*** USlargevalue -1.2823*** 0.2819*** -1.5891*** 0.1575 -1.2142** 0.1176* -2.0041*** 0.037 USmidcap -2.1359*** 0.1582** -1.9125*** 0.2656*** -2.0007*** 0.1896** -2.3070*** 0.1563** _cons 0.6169 0.9160*** -1.7682 1.6383*** 0.8999 1.1514*** -8.1729** 0.2718 N 790 852 529 562 172 182 161 170 r2 0.0003 0.007 0.0756 0.4327 0.1203 0.1195 0.2466 0.2983 This table reports the coefficients for Panel Data models for the Alpha and Sharpe performance measures after 12 and 36 months. Gold, Silver and Bronze are dummies to control the medal of a fund obtained from Morningstar Analyst Rating. ManagerTen, the manager tenure, Netexp, the net expense ratio and LogSize, the logarithm of the assets of the fund, are control variables. Finally, USlargeblend, USlargegrowth, USlargevalue and USmidcap are dummies to control for categories. N is the number of observations, r2 is a measure of the goodness of fit. *Significant at 10%; ** significant at 5% and *** significant at 1%. Chapter III: Does Morningstar Analyst Rating matters for mutual funds? 75 4.4. Results for some Pillars included in the Analyst ratings (forward looking) In this section, we have included the pillars in which the analyst ratings are broken down to analyse whether the future performance is related to any specific dimension. Specifically, we consider only three pillars, excluding Price, because the expense ratio and performance are already included as a control variable, since the quantitative ratings that contain such information are subsequently analysed. The obtained results show, in general, non-significant coefficients, with both negative and positive sign. The fact that there are very few mutual funds with a negative pillar leads to a comparison between fundamentally positive and neutral pillars. The high subjectivity that can lead to the evaluation process and the difficulty to establish a threshold between a positive and a neutral evaluation can explain the results obtained. Table 18Pillars and out-of-sample performance after 12 and 36 months Variable Alpha 12 Sharpe 12 Alpha C-12 Sharpe C-12 Alpha 36 Sharpe 36 Alpha C-36 Sharpe C-36 Parentpos 0.5001 0.0688* -0.0211 -0.0089 0.21 0.0294 -0.1735 -0.0307 Peoplepos -0.543 -0.0881 -0.1532 -0.0304 0.1531 0.0431 -0.0274 0.007 Processpos -0.284 -0.0275 0.3184 0.0293 -0.4298 -0.1147* 0.4842 0.0356 USlargeblend -2.1777*** 0.2538*** -2.8730*** 0.1849*** -2.5693*** 0.0906 -3.2014*** 0.0239 USlargegrowth -1.8044*** 0.2255*** -1.9416*** 0.2335*** -1.1939 0.1937** -1.4232** 0.1829** USlargevalue -1.1168* 0.2754*** -1.7632*** 0.1844** -1.2380* 0.1343 -2.2574*** 0.0195 USmidcap -1.5612** 0.2857*** -1.6983** 0.2659*** -1.7527** 0.1782* -1.9586** 0.1650* Netexp_ - - -0.9361* -0.1591*** - - -0.8113 -0.1491** logSize - - 0.4038** 0.0470** - - 0.6220*** 0.0736*** ManagerTen_ - - -0.0088** -0.0009** - - -0.0095* -0.0010* _cons 3.3915*** 0.3681*** 0 0 1.1183 1.1884*** -11.0519** -0.2456 N 527 565 478 493 130 130 123 123 r2 0.1586 0.8135 0.1647 0.8243 0.1219 0.0974 0.2539 0.3118 This table reports the coefficients for Panel Data models for the Alpha and Sharpe performance measures after 12 and 36 months. Parentpos, Peoplepos, Processpos are dummies which take the value 1 in case the Pillars Parent, People and Process of the Morningstar Analyst Rating is positive, 0 otherwise. ManagerTen, the manager tenure, Netexp, the net expense ratio and LogSize, the logarithm of the assets of the fund, are control variables. Finally, USlargeblend, USlargegrowth, USlargevalue and USmidcap are dummies to control for categories. N is the number of observations, r2 is a measure of the goodness of fit of the model. *Significant at 10%; ** significant at 5% and *** significant at 1%. 4.5. Results for stars ratings (backward looking) The models estimated for the stars rating display different results than the analyst, showing that the ratings of 4 outperform those considered as not recommended according to this criterion in terms of alpha and Sharpe but only in the short term (12 months) and without control variables. Given that four one-year periods are used in the panel regression, the results indicate that investing for the term of one year starting at any one of them, yields better results for investors with four stars funds. However, this superiority is not maintained in the long term, Are the ratings useful tools selecting mutual funds? 76 suggesting, as for analyst, the need to monitor the portfolio on a yearly basis and to verify that the funds are preferably rated 4 stars. Table 19Stars ratings and out-of-sample performance after 12 and 36 months Variable Alpha 12 Sharpe 12 Alpha C-12 Sharpe C-12 Alpha 36 Sharpe 36 Alpha C-36 Sharpe C-36 stars5 0.1855 0.0404 -0.3248 0.034 -0.2197 0.0227 0.1997 0.0427 stars4 0.9753** 0.0975* 0.0327 0.0544 -0.2543 -0.0016 -0.2489 -0.0073 stars3 0.1446 0.0221 -0.701 0.0166 0.2108 0.0125 0.7859 0.0703 Netexp_ - - -0.6308 -0.1121** - - -0.4743 -0.1342*** LogSize_ - - 0.2432 0.0342* - - 0.5132*** 0.0500*** ManagerTen_ - - -0.0076** -0.0009** - - -0.0085** -0.0008** USlargeblend -1.6681*** 0.2541*** -2.8737*** 0.1457** -2.3813*** 0.1524*** -3.3087*** 0.0652 USlargegrowth -2.3913*** 0.1332*** -1.9423*** 0.1703*** -1.4986*** 0.2284*** -2.1084*** 0.1770*** USlargevalue -1.1255** 0.2728*** -1.5445*** 0.1567** -1.5083*** 0.1261** -2.3011*** 0.0473 USmidcap -1.9233*** 0.1812*** -1.9599*** 0.2064*** -2.1067*** 0.2036*** -2.5243*** 0.1639** _cons -1.1438** 0.3607*** -3.2438 0 1.2343** 1.1152*** -8.5187** 0.2432 N 873 930 587 618 218 229 201 211 r2 0.1334 0.8217 0.1747 0.841 0.1238 0.0964 0.2734 0.273 This table reports the coefficients for Panel Data models for the Alpha and Sharpe performance measures after 12 and 36 months. Stars5, stars4, stars3 are dummies which reflect the number of stars of Morningstar Star Rating. ManagerTen, the manager tenure, Netexp, the net expense ratio and LogSize, the logarithm of the assets of the fund, are control variables. Finally, USlargeblend, USlargegrowth, USlargevalue and USmidcap are dummies to control for categories. N is the number of observations, r2 is a measure of the goodness of fit of the model. *Significant at 10%; ** significant at 5% and *** significant at 1%. 4.6. Results for best stars and analyst ratings (combining forward and backward looking) In this section, we want to check if the combination of both ratings can help in the process of identifying outperformers. Thus, we created different indicator variables that result from the combination of the best analyst (Gold to Bronze) and stars ratings (4 or 5 stars) and considering the rest as “Not recommended”. The results of the Table 20 show that within 12 months only the gold funds of four or five stars outperform the funds with the worst ratings. When we take the 36-month term, the differences are positive in terms of Sharpe ratio for the categories bronze4or5, gold3stars and bronze3stars. The rest of the signs are generally positive, except in the case of silver3stars, but not significant. The results of the joint analysis of the funds show that when the ratings are used in isolation, only the fouror five-star funds and gold funds outperform the poorer quantitative rating in the short term. When both criteria are combined, again, the funds that result from combining gold and four or five stars perform better for the 12month term, but not the 36-year term. This means that investors who base their decisions on both criteria must monitor portfolios annually and check that they continue to maintain both ratings. On the other hand, the combination of the two ratings does have medium-term differentiation results, with a higher performance in terms of Sharpe's ratio for bronze (3, 4 or 5 stars) or threestar gold funds, presenting only a negative sign the three-star silver backgrounds. Chapter III: Does Morningstar Analyst Rating matters for mutual funds? 77 Table 20Stars and analyst rating combination and out-of-sample performance after 12 and 36 months. Variable Alpha 12 Sharpe 12 Alpha C-12 Sharpe C-12 Alpha 36 Sharpe 36 Alpha C-36 Sharpe C-36 gold4or5 0.9131* 0.0970* 0.1859 0.0182 -0.0055 0.1124 -0.7422 -0.0224 silver4or5 0.288 0.0325 -0.6254 -0.0328 -0.4702 0.081 -0.711 0.0289 bronze4or5 0.0651 0.0369 0.0327 0.0469 -0.0156 0.1152** 0.0466 0.0598 gold3stars 0.3567 0.1063 -0.5773 -0.0386 1.2412* 0.3196** 1.0858 0.2475* silver3stars -0.445 -0.0272 -1.0164 -0.02 -0.4032 -0.1295 -0.5996 -0.1546* bronze3stars -0.1211 -0.0208 -0.799 -0.0341 1.1077 0.1938** 0.9349 0.1988** USlargeblend -1.6695*** 0.2637*** -2.9036*** 0.1426** -2.9453*** 0.5724*** -3.5135*** 0.4660*** USlargegrowth -2.3087*** 0.1507*** -2.0591*** 0.1648** -2.0367*** 0.6547*** -2.4280*** 0.5690*** USlargevalue -1.0088** 0.2855*** -1.5626*** 0.1584** -1.9786*** 0.5309*** -2.3603*** 0.4557*** USmidcap -1.8791*** 0.2019*** -2.0078*** 0.2109*** -2.2934*** 0.6345*** -2.5141*** 0.5770*** Netexp_ - - -0.6264 -0.1216** - - 0.0439 -0.0633 logSize - - 0.2281 0.0350* - - 0.4279*** 0.0520** ManagerTen_ - - -0.0070** -0.0008** - - -0.0062* -0.0002 _cons -0.8995* 0.3755*** 0 0 1.5764*** 0.6344*** -6.8484** -0.3307 N 874 942 588 627 331 484 305 406 r2 0.1254 0.8213 0.1715 0.8425 0.1979 0.3566 0.2975 0.3842 This table reports the coefficients for Panel Data models for the Alpha and Sharpe performance measures after 12 and 36 months. Gold4or5 is a dummy variable with the value of 1 if the fund is rated Gold with 4 or 5 stars, silver4or5 is a dummy variable with the value of 1 if the fund is rated Silver with 4 or 5 stars, bronze4or5 is a dummy variable with the value of 1 if the fund is rated Bronze with 4 or 5 stars, gold3stars is a dummy variable with the value of 1 if the fund is rated Gold with 3 stars and silver3stars is a dummy variable with the value of 1 if the fund is rated Silver with 3 stars. ManagerTen, the manager tenure, Netexp, the net expense ratio and LogSize, the logarithm of the assets of the fund, are control variables. Finally, USlargeblend, USlargegrowth, USlargevalue and USmidcap are dummies to control for categories. N is the number of observations, r2 is a measure of the goodness of fit of the model. *Significant at 10%; ** significant at 5% and *** significant at 1%. 5. Robustness We conducted some additional robustness tests to check the consistency of our results and to provide other complementary analysis. In addition to panel data we also use the quantile regression to extend the regression model to conditional quantiles of the different performance metrics because it is more appropriate for a heterogeneous mutual fund universe where strategies and objectives can vary (Chen and Huang ;2011). This model let us capture information about the coefficients at different quantiles of the dependent variable given the set of endogenous variables (star rating). In addition, the conditional quantile regression developed by Koenker and Bassett (1978) deals well with skewed distributions of fund performance. In particular, we adopt the bootstrapping method proposed by Efron (1979) and implemented in the software Stata 12. Given Y! as the different performance metrics used in this paper, and X! as a vector of exogenous variables representing the rating of the fund, the quantile model can be written as: 𝑦!= 𝑋! ´𝛽!+𝑢!" Are the ratings useful tools selecting mutual funds? 78 Assuming that: 𝑄𝑢𝑎𝑛𝑡!(𝑦!|𝑋!)= 𝑋! ´𝛽! 𝑄𝑢𝑎𝑛𝑡!(𝑢!"|𝑋!)=0 As can be seen in Table 21, the results of the quantile regression show that in general the signs are not significant for most of the ratings, with gold being the only ones that outperform those not recommended in quartile 25 and 75. When included the management costs, the gold ratings that are significant, cease to be, and it is therefore reasonable to think that both the size of the funds and the costs can explain the differences in performance more than analyst ratings. Table 21Quantile regression (Out of sample performance after 12 months). Variable Alpha-12 Sharpe-12 AlphaC-12 SharpeC-12 q25 Gold 0.7348 0.1542** -0.7879 0.0494 Silver -1.3122** -0.0175 -1.4756** -0.0429 Bronze -0.3791 0.0083 -0.0466 0.0386 Netexp_ - - -2.6220*** -0.1997*** logSize - - 0.2811* 0.0165 ManagerTen - - -0.0056 -0.0008 cons -3.9877*** 0.0974 -4.4653 1.2791** q50 Gold 0.1641 0.0577 -0.4749 0.0068 Silver -0.3934 0.0066 0.066 0.0096 Bronze -0.1826 -0.0113 0.3432 0.0671 Netexp_ - - -0.4713 -0.1107 logSize - - 0.4343*** 0.0399 ManagerTen - - -0.0104** -0.001 _cons -0.3282 0.4845*** -7.6554** 1.2672* q75 Gold 1.1570** 0.0809 0.8035 -0.0494 Silver 1.1729** 0.0853 1.0237 0.0723 Bronze 0.4998 0.0378 0.6718 0.0846* Netexp_ - - -0.2892 -0.1315 logSize - - 0.1479 0.0543** ManagerTen - - -0.0025 -0.0014*** _cons 2.4698*** 0.6826*** -0.6839 1.5432** N 790 852 529 562 This table reports the coefficients for Quantile regression. Alpha12 is the beta-adjusted return over a one-year period; Sharpe12 is the yearly risk-adjusted return. Gold, Silver and Bronze are dummies to control the medal of a fund obtained from Morningstar Analyst Rating. ManagerTen, the manager tenure, Netexp, the net expense ratio and LogSize, the logarithm of the assets of the fund, are control variables. N is the number of observations. *Significant at 10%; ** significant at 5% and *** significant at 1%. One reason that might explain why analyst ratings were not very significant in previous analysis may be because "Morningstar analyst rating is a qualitative, forward-looking measure that reflects the analyst's expectation of the future performance relative to its peers over a Chapter III: Does Morningstar Analyst Rating matters for mutual funds? 79 business cycle" and thus the analysis is only coherent when we compare peers, or the same categories. Thus, in this section we try to make the analysis only for large-scale categories because they have enough number of mutual funds. As can be seen in Table 22, the analysis by categories shows that gold funds outperform neutral or negative funds in the 12-month term and partially in the 36-month term in the case of the large-blend category. The same happens in the Growth category, where gold funds return again outperform to those not recommended in most of the metrics, terms and including control variables. This capacity is not observed in the case of Value, where most of the signs are positive but not significant. Therefore, the gold ratings allow identify funds that exceed their peers, mainly for a term of one year but not in all categories. Comparing the results with the main model we observe that in general for some categories we obtain that gold ratings are more significant but in general, very few differences has been obtained. Table 22Analysis for large-scale categories. Category Blend Variable Alpha-12 Sharpe-12 AlphaC-12 SharpeC-12 Alpha-36 Sharpe-36 AlphaC-36 SharpeC-36 Gold 1.1925* 0.2853** 2.1843** 0.5380*** 0.7194 -0.0231 1.6771** 0.0472 Silver -0.4759 0.0583 -0.6327 0.0966 -0.0169 -0.1272 0.0149 -0.1131 Bronze -0.3932 0.0063 -0.0636 0.078 -0.1105 -0.0896 0.8181 -0.0028 Netexp_ - - -0.4452 0.0028 - - -0.9610* -0.1046* logSize - - -0.3183 -0.0476 - - -0.0095 0.0165 ManagerTen - - -0.0218*** -0.0055*** - - -0.0240*** -0.0025*** _cons -1.1341* 1.1393*** 7.6801* 2.7090*** -1.7539*** 1.1191*** 1.2602 1.2444*** N 191 215 135 146 90 106 39 46 r2 0.0187 0.0291 0.1054 0.3565 0.0277 0.0283 0.4702 0.4784 Category Growth Variable Alpha-12 Sharpe-12 AlphaC-12 SharpeC-12 Alpha-36 Sharpe-36 AlphaC-36 SharpeC-36 Gold 2.5081*** 0.2483 2.9677** 0.4703** 2.5668** 0.1345 3.0034** 0.2017** Silver 0.1032 -0.2511* -0.9034 -0.1225 -0.3255 -0.1021 -0.274 -0.0001 Bronze 0.7343 0.0338 0.2825 0.1469 0.3727 0.0085 0.1689 0.0036 Netexp_ - - 0.0522 0.1522 - - 3.7374 0.4183** logSize - - 0.3867 -0.0268 - - 0.8891** 0.0960*** ManagerTen - - -0.0053 -0.0026 - - -0.0071 -0.0010* _cons -2.5865*** 1.0803*** -9.2674 1.8833* -1.9022*** 1.0665*** -23.0246** -1.0813 N 166 171 101 103 77 77 29 29 r2 0.0034 0.0521 0.0138 0.2426 0.1324 0.0367 0.4412 0.5076 Category Value Variable Alpha-12 Sharpe-12 AlphaC-12 SharpeC-12 Alpha-36 Sharpe-36 AlphaC-36 SharpeC-36 Gold 0.9163 0.1868 0.818 0.1301 0.1344 -0.0405 0.4533 -0.0762 Silver -0.2367 0.1761 0.0135 0.4033 -0.1647 -0.0641 0.22 -0.0588 Bronze -0.5388 -0.1056 0.9535 0.5165 -0.4786 -0.1494 0.8571 -0.0562 Netexp_ - - -3.1108*** -0.2363 - - -2.4342*** -0.1074 logSize - - 0.2379 0.0821 - - 0.1321 0.036 ManagerTen - - -0.0032 -0.0098** - - -0.0156 -0.0007 _cons -0.4915 1.1819*** -3.2229 0.0024 -0.694 1.1116*** -0.6743 0.6336 N 104 114 71 77 50 55 23 26 r2 0.0043 0.0456 0.0137 0.371 0.0132 0.0386 0.3171 0.1697 This table reports the coefficients for Panel Data models for the Alpha and Sharpe performance measures after 12 and 36 months large-blend, large growth and large value categories. ManagerTen, the manager tenure, Netexp, the net expense ratio and LogSize, the logarithm of the assets of the fund, are control variables. N is the number of observations, r2 is a measure of the goodness of fit of the model. *Significant at 10%; ** significant at 5% and *** significant at 1%. Are the ratings useful tools selecting mutual funds? 80 6. Conclusions Many investors select their investments in mutual funds based on quantitative rating. However, the selection of funds based exclusively on quantitative ratings, excludes a set of qualitative factors that can explain future performance. Morningstar has two systems to classify mutual funds based on qualitative aspects: Morningstar Analyst Rating and Morningstar Stewardship Grade. Morningstar Analyst Ratings are forward-looking qualitative and quantitative analyses of mutual fund about five pillars: Process, Performance, People, Parent and Price, that includes factor like the cost, past performance, quality of management, interest alignment, etc. Morningstar Analyst Ratings are based on the convictions that funds will outperform their benchmarks over the long term. Despite there exists several studies about quantitative ratings, Analyst Ratings is not as popular and very few research has been done about it. Kamal (2013) and Armstrong, et al. (2016) are the unique authors that focus their research on Analyst ratings. In this paper, we assess to what extent selecting mutual funds based on Morningstar Analyst and Star ratings criteria has an impact on the performance of investors. In particular, we endeavour to answer if good analyst ratings outperform non-recommended ones in the short (12 month) and long term (36 month) and if it is useful to combine both ratings in the screening process to identify good future performers. The data has been collected from Morningstar Direct database covering the period August 2012 to August 2016. We selected US equity funds with the previously mentioned ratings. Our conclusions support the ability of Gold ratings to select funds that will behave better in terms of future performance. Our results are in line with previous empirical evidence found in Kamal (2013) and Armstrong et al. (2016) but we have found little evidence that, on average, funds with a better Analyst Rating (Gold) have a better performance in terms of risk adjusted measures (alpha and Sharpe). The predictability is observed in several analyses done in one year ahead but not for three-years. This evidence is more relevant in the case of the analysis made by investment style´s category. In the analysis of if the pillars in which the analyst ratings are broken down we do not find evidence that future performance is related to any of these specific dimensions. The combination of Stars and Analyst ratings does have medium-term differentiation results, with a higher performance in terms of Sharpe's ratio for bronze (3, 4 or 5 stars) or three-star gold funds. The inclusion of other variables such as costs, size and manager tenure reflects the importance of considering other variables for fund´s selection. Nevertheless, in several estimations Gold ratings are still significant in explaining performance, indicating that costs are not the only factor that determines the predictive power of qualitative ratings. Our results support Chapter III: Does Morningstar Analyst Rating matters for mutual funds? 81 the use of qualitative ratings in the investment funds selection process, accompanied by other variables. Are the ratings useful tools selecting mutual funds? 88 similar performance to S&P 500, DSI and conventional funds. Kreander, Gray, Power and Sinclair (2002) using a matching procedure and the age, size, country and investment universe of the fund as variables. The study included mutual funds from Sweden, Netherlands, Norway, Germany, UK and Switzerland, and as performance metrics the Sharpe, Jensen`s Alpha and Treynor Ratio. Their results showed that SRI funds’ performance was very similar to those of conventional funds. Kreander, Gray, Power and Sinclair (2005) studied the performance of 30 European SRI funds from four countries findings that there is no difference between SRI funds and conventional funds. Bello (2005) studied 42 SRI U.S. mutual funds, he found no evidence of a performance difference between SRI and conventional funds. Both underperformed the Domini 400 Social Index and S&P 500 during the study period (1994 – 2001). Bauer, Koedijk and Otten (2005) investigated the performance of 32 British, 16 German and 55 U.S. SRI funds, they used Jensen and Carhart´s alpha and found that German and U.S. SRI mutual funds underperformed in both their relevant indexes and the conventional funds, whereas UK funds slightly outperformed, however the differences are not significant. Scholtens (2005) investigates the performance of Dutch SRI funds and finds that these funds outperformed conventional funds but with no statistically significant difference. Barnett and Salomon (2006) studied 61 SRI funds tracked by the US Social Investment Forum (USSIF). They found that the relationship between financial and social performance is neither strictly negative, nor strictly positive. Instead, they found a curvilinear relationship, suggesting that two viewpoints may be complementary. Riskadjusted performance varies with the types of social screens used. Community relations screening (excludes firms that do not invest in and/or develop economically depressed communities) increased financial performance, but environmental and labor relations screening (excludes firms with a record of poor environmental performance and firms with a record of poor labor relations practices, respectively) decreased financial performance. Bauer, Otten and Rad (2006) investigated the performance of Australian ethical funds, and Bauer, Derwall and Otten (2007) evidence from Canada finding no statistical difference in performance between these two types of funds. Gregory and Whittaker (2007) in the UK market found that neither SRI nor non-SRI funds exhibited significant under performance on a risk/style adjusted under any of the models. Renneboog, Ter Horst and Zhang (2008a) found that SRI funds in the US, the UK, and in many continental European and Asia-Pacific countries underperformed their domestic benchmarks. However, with the exception of France, Japan and Sweden, the risk-adjusted Chapter IV: Does Sustainability Score Impact Mutual Fund Performance? 89 performance of SRI funds is not statistically different from the performance of conventional funds. Cortez, Silva and Areal (2011) performed a study focused on 88 SRI funds from the European market from 1996 to 2007. They concluded that the performance of SRI funds is similar to the performance of both conventional and socially responsible indexes. Cortez, Silva and Areal (2012) Cortez Silva and Areal (2012) studied seven European markets and US market from 1996 to 2008. They found in several European markets (Belgium, France, Germany, Italy, the Netherlands and UK) that SRI funds showed similar performance compared to both conventional and benchmarks. In contrast, the US and Austrian funds showed evidence of underperformance. Nofsinger and Varma (2014) found that SRI mutual funds outperformed conventional funds in the global financial crisis, so they can be an optimal choice for investors who want to protect themselves from downside risk. They also found that SRI funds underperform at other times. Leite and Cortez (2014) performed a multi-country study focused on 54 international SRI funds located in eight European markets (Austria, Belgium, France, Germany, Italy, the Netherlands, UK and Spain), they applied the five-factor model and found a similar performance between socially responsible funds and conventional funds. Muñoz, Vargas and Marco (2014) studied 89 European green funds and 18 US funds, from 1994 to 2013. They applied the Carhart four-factor model and stated that for the US market, green funds did not perform any worse than the market, but with a global equity portfolio green funds showed evidence of underperformance. Becchetti, Ciciretti, Dalo and Herzel (2015) find no clear-cut dominance over the entire period (19922012) but also find that SRI funds generally did better than conventional funds in the period following the global financial crisis of 2007. Leite and Cortez (2015), focusing on the French market, found that SRI funds underperformed slightly more than their matchedsamples according to different models, but differences in alphas are not statistically significant in most cases. They only found significance in one of the estimated models at the 10% level. Humphrey, Warren and Boon (2016) found that SRI managers have longer tenure and are more likely to be female, but they did not find any significant difference in the performance of SRI and conventional funds. A recent and relevant reference is El Ghoul and Karoui (2017), to our knowledge this is the only paper that does not employ a dichotomous criterion in the selection of SRI mutual funds. Authors employ a corporate social responsibility score (CSR score), which is an asset-weighted composite CSR fund score. They showed the effects of CSR on fund performance, compared to low-CSR funds high-CSR funds displayed a poorer performance. In Table 23 we have summarized previous research on SRI fund performance. Are the ratings useful tools selecting mutual funds? 90 Chapter IV: Does Sustainability Score Impact Mutual Fund Performance? 91 Table 23.- Previous research on SRI fund performance Authors Sample (country and time spam) Fund Sample Relationship Conclusions Luther, Matatko and Corner (1992) UK; 1984-1990 15 SRI funds Positive -SRI funds outperform the index. There is clear evidence that the “ethical” trusts have UK investment portfolios more skewed towards companies with low market capitalization than the market as a whole. Hamilton, Jo and Statman (1993) US; 1981--1990 32 SRI funds Neutral -In general, the average of performance is similar between SRI funds and conventional funds. Luther and Matatko (1994) UK; 1985-1992 9 SRI Funds Neutral -No significant difference between SRI funds and conventional funds White (1995) US and Germany, 19911993 6 US funds and 5 German SRI funds Negative -SRI investments underperform the benchmark in different performance measures Mallin, Saadouni and Briston (1995) UK; 1986-1993 29 SRI Funds Neutral Positive -Ethical funds performed as well as their non-ethical counterparts and better than the non-ethical funds when the Jensen performance measure was used Gregory, Matatko and Luther (1997) UK, 1986-1994 18 SRI funds Neutral -No significant difference in financial performance exists between the two groups of funds Statman (2000) US; 1990-1998 31 SRI funds Neutral -Most of the SRI funds have a similar performance as the S&P 500-index and the DSI. SRI funds exhibit a positive but not significant Jensen´s alpha relative to the conventional funds Kreander, Gray, Power and Sinclair (2002) Belgium, Germany, Netherlands, Norway, Sweden, Switzerland, UK; 1986-1998 40 SRI Funds Neutral -On average, the SRI funds gave the same returns as conventional funds. There is some evidence that ethical funds are less risky as measured by volatility of returns and fund beta than their non-ethical counterparty. Kreander, Gray, Power and Sinclair (2005) Germany, Sweden, Netherlands and UK 1995-2001 30 SRI funds Neutral -There is no difference between ethical and non-ethical funds according to the performance measures employed Bello (2005) US; 1994 – 2001 42 SRI funds Neutral -No significant difference in investment performance between SRI and conventional funds. Bauer, Koedijk and Otten (2005) German, UK, US, 19902001 103 SRI funds Neutral -SRI mutual funds underperform the conventional funds; however, differences are not statistically significant Scholtens (2005) Netherlands, 2001-2003 12 SRI funds Neutral -SRI funds outperform conventional funds but with no statistically significant difference. Barnett and Salomon (2006) US, 1972-2000 61 SRI funds Negative/ Positive -Community relations screening increased financial performance. Environmental and labour relations screening decreased financial performance Bauer, Otten and Rad (2006) Australia, 1992-2003 25 SRI funds Negative Neutral Positive -During 1992-1996 domestic ethical funds underperformed conventional funds while international ethical funds outperformed conventional funds. During 1996-1999 domestic ethical funds outperformed the performance of conventional funds. No statistically significant difference between SRI funds and regular funds was found during 1999-2003 or in the whole period (1992-2003). Continue on next page Table 23 (Cont.).- Previous research on SRI fund performance Are the ratings useful tools selecting mutual funds? 92 Authors Sample (country and time spam) Fund Sample Relationship Conclusions Bauer, Derwall and Otten (2007) Canada, 1994-2003 8 SRI funds Neutral -Ethical funds underperformed conventional funds, but the performance differential is statistically insignificant Gregory and Whittaker (2007) UK, 1989-2002 32 SRI funds Neutral -No significant difference in investment performance between -- SRI and conventional funds. Renneboog, Ter Horst and Zhang (2008a) 23 countries and offshore jurisdiction (US, UK, continental European and Asia-Pacific countries), 1991 2003 463 SRI funds Negative Neutral -The average SRI fund in most European and AsiaPacific countries strongly underperform their benchmark portfolios. -In UK and US, the riskadjusted returns of SRI funds are not significantly different from conventional funds. Cortez, Silva and Areal (2011) Austria, Belgium, France, Germany, Italy, the Netherlands and UK, 1996 to 2007 88 SRI funds. Index benchmarks Neutral -Performance of SRI funds is similar to the performance of conventional benchmarks (MSCI AC World Index and MSCI AC Europe Index) and socially responsible benchmarks (FTSE4Good Global Index and FTSE4Good Europe Index) Cortez, Silva and Areal (2012) European market (Austria, Belgium, France, Germany, Italy, the Netherlands and UK) and US market, 19962008 39 funds for European markets and 7 funds for US market Negative Neutral -SRI funds for Belgium, France, Germany, Italy, the Netherlands and UK show similar performance compared to both conventional and benchmarks. -In contrast, for US and Austrian funds they showed evidence of underperformance Nofsinger and Varma (2014) US, 2000-2011 240 SRI funds Neutral Positive -Alphas for the SRI funds are not significantly different than conventional fund alphas. -It is a slight insignificant underperformance compared to conventional funds during the noncrisis periods. During the crisis periods (arch 2000 to October 2002 and October 2007 to March 2009), the SRI funds outperformed the conventional funds. Leite and Cortez (2014) Austria, Belgium, France, Germany, Italy, the Netherlands, UK and Spain, 20002008 54 SRI funds Neutral -Differences in the performance of international SRI funds and their conventional peers are not statistically significant Muñoz, Vargas y Marco (2014) US and European countries, 1994-2013 18 US and 89 European green funds Negative Neutral -For US SRI funds obtain statistically significant performance in crisis periods but underperform relative to the market in non-crisis periods. For European SRI funds obtain statistically insignificant performance irrespective of market conditions. Leite and Cortez (2015) France, 2000-2008 50 SRI funds Negative Neutral -SRI funds underperformed slightly more than their matched-samples according to different models, but differences in alphas are not statistically significant in most cases Becchetti, Ciciretti, Dalo and. Herzel, (2015) Global,1992–2012 1,213 SRI funds Positive -Socially responsible funds played an outperformed conventional funds during the 2007 global financial crisis El Ghoul and Karoui (2017)) US, 2003-2011 2,168 funds Negative -The CSR score of the portfolio is negatively related to risk-adjusted performance. Chapter IV: Does Sustainability Score Impact Mutual Fund Performance? 93 3. Background of Sustainalytics´ Methodology and Morningstar 3.1 Sustainability Scores. 3.1.1. Sustainalytics´ Methodology Sustainalytics is a global company leader in ESG research and analysis. Sustainalytics´ ESG methodology consists of approximately 150 ESG indicators7 to measure a company’s sustainable practices. Sustainalytics assess company performance based on several internal and external data sources (Sustainalytics, 2016): review of company reporting (annual reports, etc.), review of external sources (NGOs, publications, etc.), analysis is done by an experienced analyst, structural peer review, company feedback and research process. The three pillars of the ESG Score are Environment, Social, and Governance. In each pillar several categories are distinguished (Figure1), for example in the Pillar Environment you can see Operations, Supply channel and Products and Services. Within these categories there are various indicators. Figure 1Sustainalytics Framework. Source: Sustainalytics Framework. Sustainalytics use two kinds of indicators templates: core and sector-specific. Core indicators are those used for all companies. Sector-specific indicators are those used to adjust to sector specific characteristics when it is not considered. There are two main types of scoring schemes for indicators: binary and linear. Binary indicators are those which the possible raw scores for binary indicators are 0 or 100. Linear indicators: are those which 7 The exact number depends on the company´s industry. Environmental! Opera/ons! Supply!channel! Products!&!Services! Social! Employees! Supply!channel! Customers! Community!&! Philanthropy! Governance! Business!Ethics! Corporate! Governance! Public!Policy! Are the ratings useful tools selecting mutual funds? 94 there are various possible raw scores (0, 25, 50, 75 or 100). For the ESG indicators, Sustainalytics differentiate between three types that focus on different dimensions: preparedness, disclosure and performance (Sustainalytics, 2016): • Preparedness indicators assess if company management systems and policies are well designed to manage material ESG risks. • Disclosure indicators assess if a company reporting meets International best practice standards and if is transparent with respect to most material ESG issues. • Performance indicators assess ESG performance based on quantitative metrics (for example carbon intensity) and qualitative based on the analysis of controversial incidents. There are special indicators that assess whether companies are involved in certain controversies. Controversies fall into five categories: category 1 – low, 2 – moderate, 3 – significant, 4 – high and 5 – severe. Controversy topics include: Business Ethics, Society and Community, Environmental Operations, Environmental Supply Chain, Product and Service, Employee, Social Supply Chain, Customer, Governance, and Public Policy. Finally, to calculate the total score of the company, as well as aggregate scores on the three pillars, Sustainalytics uses a default weight matrix that is uniquely defined for every peer industry group (42 different comparable sub-industries). 3.1.2. Morningstar Sustainability and ESG Scores The Morningstar Portfolio Sustainability Score is a measure developed in 2016 for scoring mutual funds and ETFs about environmental, social, and governance, or ESG, risks and opportunities. The subsequent Morningstar Sustainability Rating is a comparison relative to their Morningstar Category peers and is derived from the Morningstar Portfolio Sustainability Score. Morningstar Portfolio Sustainability Score (Sustscore) is defined as follows (Morningstar, 2016a and 2016b): 𝑆𝑢𝑠𝑡𝑆𝑐𝑜𝑟𝑒 = 𝑃𝑜𝑟𝑡𝑓𝑜𝑙𝑖𝑜 𝐸𝑆𝐺 𝑆𝑐𝑜𝑟𝑒 − 𝑃𝑜𝑟𝑡𝑓𝑜𝑙𝑖𝑜 𝐶𝑜𝑛𝑡𝑟𝑜𝑣𝑒𝑟𝑠𝑦 𝐷𝑒𝑑𝑢𝑐𝑡𝑖𝑜𝑛 In order to receive a portfolio sustainability score, a portfolio must have a portfolio ESG score and a portfolio controversy score, which according to Morningstar (2016b) at least of 50% of a portfolio’s assets under management must have a company ESG score and a controversy score/deduction. Based on their portfolio Sustainability score, ESG scores and Chapter IV: Does Sustainability Score Impact Mutual Fund Performance? 95 controversy scores, and descriptive ranks within their Morningstar Categories, funds are ranked as 5 (High), 4 (Above Average), 3 (Average), 2 (Below Average) and 1 (Low) (see Table 24 and 25). Table 24Portfolio Sustainability score and Sustainability Rating. Distribution Score Descriptive Rank Highest 10% (best) 5 High Next 22.5% 4 Above Average Next 35% 3 Average Next 22.5% 2 Below Average Lowest (worst) 1 Low Source: Morningstar (2016b). Table 25-Portfolio Controversy Scores Distribution Score Descriptive Rank Lowest 10% (Best) 5 High Next 22.5% 4 Above Average Next 35% 3 Average Next 22.5% 2 Below Average Highest 10% (Worst) 1 Low Source: Morningstar (2016b). Morningstar Portfolio ESG Score(ESGscore)8 is calculated as: ESGscore =w!ESGNorm! ! !!! Where: ESGNorm!= the normalized ESG score of company 𝑖, 𝑛= the number of securities in the portfolio, w!= the asset weight on security 𝑖, so the sum w!=100% ! !!!. To make the ESG scores comparable across peer groups, Morningstar normalizes the scores using a z-score transformation: 𝑍!=!"#!!! ! Where: ESGi = ESG score of company 𝑖, µ = the mean of the ESG scores of the companies in the peer group, δ = the standard deviation of the ESG scores of the companies in the peer group. 8 Morningstar Portfolio Environmental Score, Social Score and Governance Score is calculated as an assetweighted average of the scores of the individual companies. Are the ratings useful tools selecting mutual funds? 96 𝑍𝑖 are used to create the normalized ESG scores9 on a 0-100 scale, with a mean of 50, as: ESGNorm!=50 +10𝑍𝑖 Normalized company ESG are aggregated to a portfolio ESG score using an assetweighted average of all covered securities. Sustainalytics tracks and categorizes ESG related incidents in companies, which is called “controversies”. A single company may be involved in multiple ESG-related incidents at any given time, so Sustainalytics makes a controversy score of company i. Morningstar employs company controversy scores of Sustainalytics creating a Morningstar Portfolio Controversy Score (MContr!), as follow: MContr!=w!SCont! ! !!! Where: w!= the asset weight on security 𝑖 SCont! = the Sustainalytics controversy score of company 𝑖. 4. Empirical study 4.1. Sample Our sample contains 1,593 European equity funds rated by Morningstar in November 2016. The funds are the "open funds" type with ESG score in the investment area of Europe. Investment area identifies the geographic area where the fund focuses their investments. Furthermore, to avoid problems of multicollinearity, we have selected only an equivalent class for each fund. We obtained for each equity mutual fund three different measures of performance and other variables such as size, volatility, social conscious, expenses and age. We also use the Morningstar style-box to control the effect of the different categories which are included in the sample. The number of funds varied when we consider the costs where the sample reduces from 1,593 to 571 motivated for the lack of data in Morningstar Direct. 9 Normalized company ESG scores can be interpreted as follows: (70-100) Company scores at least two standard deviations above average in its peer group (60-70) Company scores one standard deviation above average in its peer group 50 Company scores at peer group average (30-40) Company scores one standard deviation below average in its peer group (0-30) Company scores at least two standard deviations below average in its peer group Chapter IV: Does Sustainability Score Impact Mutual Fund Performance? 97 4.2. Variables construction Our sustainable variables have been obtained from Morningstar Direct, with original source of individual companies from Sustainalytics. We will employ five variables: three are the pillars scores [Environment score variable (Envscore), Social score variable (Socscore) and Government score variable (Govscore)], the fourth is the ESG score of a portfolio (ESGscore), and finally, the Portfolio Sustainability Score (Sustscore) which is the ESG score minus Portfolio Controversy Score. We have divided the funds into two groups based on whether ESG scores are below or above the median. Then, we estimated the means and their differences between both groups. Table 26 reports the results of the univariate analysis. As can be observed, the differences are very significant between the two groups for the different scores, with a difference of approximately five points in favour of the funds included in the high score group. Table 27 compares the funds declared sustainable (Socially conscious) and those that result from dividing the sample according to a low ESGscore or a high ESGscore criteria. As you can see, there are big differences but in general, mutual funds that are declared sustainable are from an ESGscore point of view too. However, there are many funds that are not declared sustainable but they are based on the level of sustainability of the companies that integrate the portfolio. Thus, by using scores investors have at their disposal a large number of funds that are not declared sustainable but that their portfolio is comparable to that of sustainable funds. Table 26Sustainability and ESG scores for different groups Variable Low ESG score High ESG score t-statistic Sustscore 52.63 58.58 -39.29*** ESGscore 57.44 63.99 -33.78*** EnvScore 56.66 62.36 -30.70*** Socscore 57.01 62.95 -34.81*** GovScore 55.28 60.41 -31.97*** This table reports the values of sustainability variables considered in the analysis obtained from Morningstar direct database. The funds are classified into low or high groups depending on whether their score is above or below the median. The t-statistic for difference of means is reported in the third column. Sustscore is the level of sustainability of the mutual fund measured by Morningstar. ESGscore is the ESG score of a fund. EnvScore, Socscore and GovScore are the mutual fund scores for the three dimensions (environment, social and corporate governance). *Significant at 10%; ** significant at 5% and *** significant at 1%. Are the ratings useful tools selecting mutual funds? 104 Table 33ESG Pillars models and fund performance. Variable Return2y Return1y Sharpe2y Sharpe1y Alphacat2y Alphacat1y Environment Score Models Envscore -0.3373*** -0.7500*** -0.0169*** -0.0343*** -0.0747** 0.1474** logSize 0.1717** -0.1821 0.0114** -0.0086 0.2491*** 0.1248 Sociallyconcious 1.6921** 1.1247 0.1058** 0.0621 0.5883 -0.4186 Age 0.0082 0.0492 0.0004 0.0027 -0.0026 0.0363 LossDev -0.6834*** -1.3802*** -0.0296*** -0.0680*** -0.5195*** -0.6579*** ExpRatio -0.3602 -0.4927 -0.0235 -0.0376* -0.3068 -1.2899*** _cons 37.2674*** 72.6572*** 1.4858*** 3.2394*** 6.8766** 8.683 N 571 570 571 570 541 540 r2 0.3791 0.5443 0.3379 0.4834 0.1239 0.2283 Social Score Models Socscore -0.3071*** -0.7310*** -0.0151*** -0.0335*** -0.0942*** 0.1341** logSize 0.1559* -0.2072 0.0106* -0.0098 0.2445*** 0.1271 Sociallyconcious 1.7537** 1.4089 0.1081** 0.0751 0.6708 -0.4506 Age 0.0094 0.053 0.0005 0.0029 -0.0015 0.0356 LossDev -0.7240*** -1.4174*** -0.0316*** -0.0697*** -0.5324*** -0.6445*** ExpRatio -0.3423 -0.4659 -0.0225 -0.0363 -0.3108 -1.3014*** _cons 36.9893*** 73.9977*** 1.4561*** 3.3005*** 8.3183*** 8.935 N 571 570 571 570 541 540 r2 0.3644 0.5429 0.3252 0.4823 0.1291 0.227 Government Score Models Socscore -0.1133*** 0.1867*** 0.0012 -0.0318*** 0.0012 -0.0318*** logSize 0.1907*** 0.02 0.0139*** -0.011 0.0139*** -0.011 Sociallyconcious 0.0874 -0.1171 0.0507 0.0665 0.0507 0.0665 Age -0.0244** 0.0003 -0.0007 0.0033 -0.0007 0.0033 LossDev -0.5027*** -0.7989*** -0.0382*** -0.0712*** -0.0382*** -0.0712*** ExpRatio - - -0.0248* -0.0423* -0.0248* -0.0423* _cons 10.3321*** 7.5867 0.1834 3.1898*** 0.1834 3.1898*** N 723 729 571 570 571 570 r2 0.1128 0.2555 0.3358 0.4698 0.3358 0.4698 This table reports the coefficients for the regression models for different performance measures. Alpha is the beta-adjusted return over a one-year period; Sharpe is the yearly risk-adjusted return and, Return is the total net return. Sociallyconcious is a dummy variable used to analyse sociallyconscious mutual funds. N is the number of observations and r2 the R-squared fit measure. *Significant at 10%; ** significant at 5% and *** significant at 1%. 4.8. Downside risk and sustainability scores In this part, we test if the degree of sustainability measured through ESG scores and their components, has a positive or negative effect on the historical value at risk of the portfolio. We used the following model: 𝑉𝑎𝑅!=𝑐𝑜𝑛𝑠!+𝛽!𝑆𝑢𝑠𝑡𝑠𝑐𝑜𝑟𝑒!+𝛽!𝐴𝑔𝑒!+𝛽!𝐿𝑜𝑠𝑠𝐷𝑒𝑣!+𝛽!𝑙𝑜𝑔𝑆𝐼𝑍𝐸!+ 𝛽!𝐸𝑥𝑝𝑅𝑎𝑡!+𝛽!𝑆𝑜𝑐𝑅𝑒𝑠𝑝𝑑𝑢𝑚 +𝐶𝑎𝑡𝑒𝑔𝑜𝑟𝑦!!+𝜀! As Table 34 shows, the downside risk of mutual funds is affected by the level of sustainability (ESG score). Specifically, we observed how the variable Sustscore is Chapter IV: Does Sustainability Score Impact Mutual Fund Performance? 105 negatively and significantly related with the VaR of the fund at 99% of confidence level in both terms, one and two years. These results support that funds with a higher degree of sustainability protect investors better against extreme losses. As Kurtz (1997) or Goldreyer and Diltz (1999) explain, SRI mutual funds managers base their decisions on a deeper, more complete and higher quality information, resulting in a significant reduction in the risk of their investment decisions. On the other hand, the dichotomous variable commonly used has a positive and opposite sign to that resulting from using a continuous variable. We also made the analysis for the different subfactors, observing again a negative and significant relationship for most of the estimated models. As can be seen in Table 12, the increase in the level of environmental, social and governance sustainability reduces the level of extreme losses of investment funds. It is again observed that the dummy variable is significant and positively related to the level of risk. From this analysis, we observed that the results of evaluating the effect of sustainability based on dichotomous variables may yield contradictory results to those obtained when continuous variables are used. Table 34Sustainable score and downside risk Variable VaR-2y VaR-1y VaR-2y VaR-1y VaR-2y VaR-1y VaR-2y VaR-1y Sustscore e -0.0211* -0.0277*** - - - - - - EnvScore - - -0.0193 -0.013 - - - - Socscore - - - - -0.0256** -0.016 GovScore - - - - - - -0.0289** -0.0148 logSize -0.0231 -0.0278 -0.0236 -0.0278 -0.025 -0.0285 -0.0258 -0.0291 Sociallyconcious 0.1671* 0.3396** 0.1397 0.2730* 0.1637* 0.2872* 0.1698* 0.2820* Age 0.0083** 0.0149*** 0.0082** 0.0148*** 0.0085** 0.0149*** 0.0089** 0.0151*** LossDev 0.7757*** 0.6817*** 0.7805*** 0.6745*** 0.7777*** 0.6753*** 0.7798*** 0.6744*** ExpRatio 0.0185 0.0472 0.018 0.0536 0.0163 0.0526 0.0094 0.05 Largeblend -1.1074*** -0.5057 -1.0197*** -0.5082 -1.0188*** -0.5148 -0.9981*** -0.5246 Largegrowth -1.0165*** -0.6747* -0.9683*** -0.7100* -0.9533*** -0.7043* -0.9541*** -0.7232* Largevalue -1.2509*** -0.5573* -1.1523*** -0.5113 -1.1715*** -0.5302 -1.1336*** -0.5225 Midblend -0.7316*** -0.538 -0.7191*** -0.5515 -0.7328*** -0.5607 -0.7162*** -0.5559 Midgrowth -0.6022*** -0.7343** -0.6060** -0.7415* -0.6009*** -0.7383** -0.5998*** -0.7388** Midvalue -1.0851*** -0.7785** -1.0145*** -0.6938 -1.0971*** -0.7441* -1.0475*** -0.7105* cons 2.4437*** 2.7915*** 2.3131** 2.0917*** 2.7512*** 2.2913*** 2.8750*** 2.2177*** N 571 570 571 570 571 570 571 570 r2 0.72 0.85 0.72 0.85 0.72 0.85 0.72 0.85 This table reports the coefficients for the regression models. VaR is the maximum loss that a fund i can obtain for a given time period and a given confidence level. Sustscore is the level of sustainability of the mutual fund measured by Morningstar. ESGscore is the ESG score of a fund. EnvScore, Socscore and GovScore are the mutual fund scores for the three dimensions (environment, social and corporate governance). Sociallyconcious is a dummy variable used to analyse Sociallyconcious mutual funds. N is the number of observations and r2 the R-squared fit measure. *Significant at 10%; ** significant at 5% and *** significant at 1% Are the ratings useful tools selecting mutual funds? 106 4.9. Flows and sustainability scores In this section, we analyze the effect of sustainability on the flows of investment funds. In particular, flows of sustainable funds are generally considered to be less sensitive to changes in performance because investors value other elements in their utility function. Benson and Humphrey (2008) and Renneboog et al. (2011) obtain evidence in favor of greater stability in flows for sustainable funds, while Bollen (2007) finds that SRI mutual funds are more sensitive to positive returns and less to negative ones. In line with El Ghoul and Karoui (2017) we argue that funds with higher ESG scores attract more conscious investors less worried about performance and therefore the flows are less sensitive to past performance. Thus, we estimate the following model to evaluate the effect of sustainability on the flow of funds using the different performance metrics (alpha, Sharpe, return), the sustainability score and the interaction of the product (SustPerf: sustsharpe, sustalpha or sustreturn): 𝐹𝑙𝑜𝑤!=𝛼!+ 𝛽!𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒!+𝛽!𝑆𝑢𝑠𝑡𝑠𝑐𝑜𝑟𝑒!+𝛽!𝑆𝑢𝑠𝑡𝑃𝑒𝑟𝑓 !+𝛽!𝐴𝑔𝑒!+𝛽!𝐿𝑜𝑠𝑠𝐷𝑒𝑣! +𝛽!𝑙𝑜𝑔𝑆𝐼𝑍𝐸!+𝛽!𝐸𝑥𝑝𝑅𝑎𝑡𝑖𝑜!+𝛽!𝑆ociallyconcious +𝐶𝑎𝑡𝑒𝑔𝑜𝑟𝑦! ! +𝜀! Where: SustPerf: is the product of Sustscore and sharpe (sustsharpe), alpha (sustalpha) or net return (sustreturn) depending on the model. Table 35 shows that only the model that takes as performance variable the profitability without risk adjustment, is significant. This would indicate that unadjusted returns are the ones that have had the most influence on investment decisions. On the other hand, in model three, the sustainability score is also significant, so that higher-rated funds received a larger volume of funds than those with a lower score. This fact shows that the degree of sustainability stimulates fund raising and more when the degree of sustainability is higher. Also, when we analyze the effect of the sustainability dummy variable (Sociallyconcious) it is significant in all models, which confirms the importance of sustainability in attracting investors interested in funds that are declared sustainable. This fact can be related to both greater social awareness and expectations of greater profitability in socially responsible investments. Finally, the negative sign of the interaction variable (sustreturn) shows the lower sensitivity of sustainable funds, supporting the results found by El Ghoul and Karoui (2017) using alternative metrics and US funds. Chapter IV: Does Sustainability Score Impact Mutual Fund Performance? 107 Table 35Sustainable score and flow of funds Variable Model1 Model2 Model3 Sharpe-2y 0.3936 - - Alpha-2y - 0.0143 - Return-2y - - 0.0224*** sustsharpe -0.0029 - - sustalpha - -0.0001 - susreturn - - -0.0002** Sustscore 0.0028 0.002 0.0090*** Sociallyconcious 0.1014*** 0.1129*** 0.1045*** logSize 0.0243** 0.0221** 0.0295*** Age 0.0001 0.0006 -0.0004 LossDev -0.0148** -0.0158*** -0.0056 ExpRatio 0.04 0.0382 0.0438* Largeblend -0.2039** -0.3165*** -0.1727 Largegrowth -0.2269** -0.3007*** -0.1537 Largevalue -0.1957** -0.2704*** -0.1977 Midblend -0.2237** -0.2511*** -0.1624 Midgrowth -0.2316*** -0.2645*** -0.1222 Midvalue -0.1913 -0.2839*** -0.1823 cons -0.4622* -0.2827 -1.0624*** N 560 531 560 r2 0.084 0.0709 0.0995 This table reports the coefficients for the regression models. Alpha is the beta-adjusted return over a two-years period; Sharpe is the yearly risk-adjusted return and, Return is the total net return. Sustscore is the level of sustainability of the mutual fund measured by Morningstar. Sociallyconcious is a dummy variable used to analyse socially conscious mutual funds. N is the number of observations and r2 the R-squared fit measure. *Significant at 10%; ** significant at 5% and *** significant at 1%. 5. Robustness We conducted some additional robustness tests to check the consistency of our results and to provide other complementary analysis. We checked whether performance may differ attending to the fund manager skills considering the quantiles of different performance measures: differences in the quantiles would indicate differences in the fund manager’s abilities to deal with performance. Quantile regression let capture information about the coefficients at different quantiles of the dependent variable given the set of endogenous variables. In addition, the conditional quantile regression developed by Koenker and Bassett (1978) deals well with skewed distributions of fund performance. In particular, we adopted the bootstrapping method proposed by Efron (1979) and implemented in the software Stata 12. Given 𝑌 ! as the different performance metrics used in this paper, and 𝑋! as a vector of exogenous variables representing the sustainable score of each mutual funds and other controls, the quantile model can be written as: Are the ratings useful tools selecting mutual funds? 108 𝑦!= 𝑋! ´𝛽!+𝑢!" Assuming that: 𝑄𝑢𝑎𝑛𝑡!(𝑦!|𝑋!)= 𝑋! ´𝛽! 𝑄𝑢𝑎𝑛𝑡!(𝑢!"|𝑋!)=0 Table 36 reports quantile parameter estimates for three different adjusted risk-return performances. Our evidence for all quantiles confirms no differences in the results and sustainability seems to be important independent of the level of performance analysed. We also calculated the models excluding the expense ratio because this variable has many blanks and reduces the sample a lot. After the calculations, we again observed no differences with the models presented in the previous empirical analysis. Finally, we recalculated the models for each category and we obtained different results depending on the category, concluding that on average the effect is negative on performance but specific for each category. Chapter IV: Does Sustainability Score Impact Mutual Fund Performance? 109 Table 36Quantile regression. Return2y Return1y Sharpe2y Sharpe1y Alphacat2y Alphacat1y q25 Sustscore -0.1842*** -0.3622*** -0.0115*** -0.0205*** -0,0046 0.1326** logSize 0.1722* -0,1016 0,0086 -0,01 0.2409** -0,0609 Sociallyconcious 0,999 0,4466 0,06 0,0575 0,2995 0,0173 Age 0,0149 0,0525 0,0008 0.0039* 0,0044 0.0675* LossDev -0.7367*** -1.1443*** -0.0234** -0.0511*** -0.4590*** -0.6372*** ExpRatio -0.8319*** -0.9671*** -0.0515** -0.0706*** -1.1022*** -1.7055*** cons 30.6208*** 39.3289*** 1.0995*** 2.0197*** 3,7736 4,2163 q50 Sustscore -0.2361*** -0.6639*** -0.0131*** -0.0356*** -0.0687*** 0,077 logSize 0.1673** -0,0868 0,0089 -0,0079 0.2665*** -0,032 Sociallyconcious 1,0751 1.8932* 0,0815 0.1372** 0,7089 -0,1515 Age 0,0103 0,0626 0,0007 0,0043 -0,0109 0,0347 LossDev -0.4361*** -1.2064*** -0.0204*** -0.0631*** -0.3271*** -0.6350*** ExpRatio -0.6436** -0.7434* -0.0364** -0.0661*** -0.4729** -1.6412*** cons 30.4631*** 57.1273*** 1.3114*** 3.0006*** 4,6984 11,0889 q75 Sustscore -0.3204*** -0.8217*** -0.0157*** -0.0365*** -0.0992*** -0,0348 logSize 0,0408 -0,0436 -0,0024 0,0105 0,0465 -0,0804 Sociallyconcious 1.7210** 2,3816 0.1026* 0,0727 0,9631 -0,0929 Age -0,0272 0.0826* -0,0014 0,0042 -0,0083 0,0065 LossDev -0.6449*** -1.5050*** -0.0331*** -0.0803*** -0.3354*** -0.4743*** ExpRatio -0,442 0,3107 -0,0344 0,0134 -0.5127* -1.1507*** cons 39.7659*** 83.9232*** 1.7920*** 3.6249*** 10.7921*** 27.0078** N 571 570 571 570 541 541 This table reports the coefficients for the quantile regression models. Sustscore is the level of sustainability of the mutual fund measured by Morningstar. Sociallyconcious is a dummy variable used to analyse socially conscious mutual funds. N is the number of observations. *Significant at 10%; ** significant at 5% and *** significant at 1%. Apenddix 111 6. Conclusion Socially Responsible Investment (SRI) summaries any investment strategy which search for a financial return and encourage corporate practices that promote environmental care, consumer protection and human rights. In Europe, SRI strategies grew by 11.7% from 2014 to 2016 to reach $12.04 trillion. Traditional studies focus their work on mutual funds which declare themselves as funds that support a SRI approach. One important limitation of this approach is that results could be biased, because SRI mutual funds could have different levels of sustainability and differences with conventional funds could be not significant. Recently, Morningstar launched Morningstar Sustainability Score to classifying mutual funds about ESG factors. The use of sustainability scores in our work can allow us to evaluate the effect of the degree of sustainability on performance, risk or flows on European equity mutual funds. Our result shows that there are a large number of funds that are not declared sustainable but their portfolio is comparable to sustainable mutual funds. Furthermore, Sustainability score is significant explaining the level of performance for all the metrics analysed (alpha, Sharpe and net return), with negative sign in most models. Using a conventional dummy to declare social mutual funds, the results are significant but with the contrary sign, showing that considering the level of sustainability can help to understand better the link between performance and social responsibility. Our results are in accordance to Statman and Glushkov (2016), who concluded that the lack of clearly defined criteria to distinguish SRI mutual funds affected the results. Also, we obtained similar results to El Ghoul and Karoui (2017) for US mutual funds market. Using the different pillars of ESG scores (environmental, social, and governance) we were able to achieve a negative link between the dimensions of sustainability and performance, showing that all the dimensions play an important role in explaining performance. In terms of downside risk, the level of sustainability is negatively and significantly related to the VaR of the fund, supporting that higher scored mutual funds protect better against extreme losses. The opposite is found for the conventional dummy, showing the advantages of employing a quantitative measure of sustainability to evaluate assets´ risk. This result could mean that SRI mutual funds managers base their decisions on a deeper analyse resulting in a significant reduction in the risk of their investment decisions. Our work shows that sustainability scores can be used by investors worried by extreme losses and not only by values-motivated investors. Finally, we analyzed the effect of sustainability on the flows realizing that unadjusted returns have the most influence on investment decisions. The sustainability score is significant on the flows, so higher-rated funds received a larger volume of funds and it is also significant the Are the ratings useful tools selecting mutual funds? 112 effect of the sustainability dummy variable. Finally, the negative sign of the interaction variable (product of sustainability and return) shows the lower sensitivity of sustainable funds. This shows the different sensitivity to performance of values-motivated investors. The limitations of our work are that we do not have a panel database, so we only have the observed values of sustainability and ESG scores on a data point (December 2016). Future research could use panel data and mix the Morningstar ESG scores with the MSCI ESG scores to analyze the effect of several scores. Apenddix 113 Chapter V. Conclusions, limitations and future research Mutual funds are a product highly requested by investors and savers. Our work seeks to find aspects that can predict the selection of the best funds, more specifically, we intend to look at the extent to which quantitative, qualitative and social/environmental ratings can help to disentangle the funds that will have a better performance in the future, as well as preserve longterm wealth. Choosing not only the funds that get better risk adjusted returns, but also those that are more robust in adverse times. In this study, the Chapter II focuses on quantitative ratings (Star Ratings), the Chapter III focuses on the qualitative ratings (Analyst Ratings) and finally, the Chapter IV focuses on socially responsible scores. All chapters, are based in Morningstar ratings because it is the most important mutual fund research firm and of important matter to the investors, savers, financial institutions and academics, for its independence and specialization in mutual funds. Morningstar quantitative ratings, which evaluate funds with stars as do hotels (1 to 5), has become a very popular tool for investors. Qualitative and Sustainability ratings are not yet as widely used by the fund industry as they are relatively recent, especially the latter. Many investors select their investments in mutual funds based exclusively on the quantitative rating. In this sense, we wanted to verify if the exclusive use of this criterion that allows making good decisions related to performance and downside risk. We selected European equity funds, we used Rating Overall and also three years ratings to check the robustness of our estimates. Our conclusions supports the ability of quantitative ratings to select funds that will behave better in terms of future performance. The results are in line with some previous empirical evidence found in Morey and Gottesman (2006), Müller and Weber (2014), and Meinhardt (2014). Consequently, we have found that on average, funds with a lower rating have a worse performance in terms of risk adjusted measures and Annual Return for the following year. The strongest predictability is observed one year ahead but it is also good for three-years. The inclusion of other variables such as costs, size and age reflects the importance of considering other variables for funds selection. Nevertheless, the ratings are still significant in explaining performance, indicating that costs are not the only factor that determines the predictive power of quantitative ratings. Finally, our analysis in terms of VaR (value at risk) shows that the investment in good rated funds can help to preserve the investors wealth better. Are the ratings useful tools selecting mutual funds? 120 Cortez, M.C., Silva, F. & Areal, N. (2012). Socially responsibe investing in the global market: The performance of US and European funds. International. Journal of Finance and Economics, 17(3), 254–271. Cremers, M., Driessen, J., Maenhout, P. & Weinbaum, D. (2009). Does skin in the game matter? Director incentives and governance in the mutual fund industry. Journal of Financial and Quantitative Analysis, 44(6), 1345-1373. Das, P. K. & Uma Rao, S.P. (2014). Performance persistence in socially responsible mutual funds. International Journal of Economics and Business Research, 8(4), 490–503. Del Guercio, D., Dann, L. Y. & Partch, M. M. (2003). Governance and boards of directors in closed-end investment companies. Journal of Financial Economics, 69(1), 111-152. Del Guercio, D., & Tack, P. A. (2008). Star Power: The Effect of Morningstar Ratings on Mutual Fund Flow. Journal of Financial and Quantitative Analysis, 43(4), 907-936. DiBartolomeo, D. & L. Kurtz, 1999. Managing Risk Exposures of Socially Screened Portfolios, Northfield Information Services, Boston, Available at www.northinfo.com. Ding, B. & Wermers, R. (2012). Mutual Fund Performance and Governance Structure: The Role of Portfolio Managers and Boards of Directors. Available at SSRN: https://ssrn.com/abstract=2207229 Duret, A.-S., Hereil, P., Mitaine, P., Moussavi, N., & Roncalli, T. (2008). Fund rating systems and performance predictability. Working Paper. University of Evry. Available at SSRN: http://ssrn.com/abstract=1121383. El Ghoul, S. & A. Karoui (2017). Does corporate social responsibility affect mutual fund performance and flows?. Journal of Banking and Finance, 77, 53-63. Efron, B. (1979). Bootstrap methods: another look at the jackknife. Annals of Statistics, 7(1), 1-26. Evans, A. L. (2008). Portfolio manager ownership and mutual fund performance. Financial Management, 37(3), 513–534. Faff, R., Parwada, J., & Poh, H. (2007). The Information Content of Australian Managed Fund Ratings. Journal of Business Finance & Accounting, 34 (9-10), 1528-1547. Ferson, W., & Schadt, R.W. (1996). Measuring Fund Strategy and Performance in Changing Economic Conditions. The Journal of Finance, 51(2), 425-461. Flammer, C. (2015). Does Corporate Social Responsibility Lead to Superior Financial Performance? A Regression Discontinuity Approach. Management Science, 61(11), 2549-2568. Friede, G, Busch, T. & Bassen, A. (2015). ESG and financial performance: aggregated evidence from more than 2000 empirical studies, Journal of Sustainable Finance & Investment, 5(4), 210–233. Apenddix 121 Ferris, S. P. & Xuemin, Y.S. (2007). Do independent directors and chairmen really matter? The role of boards of directors in mutual fund governance. Journal of Corporate Finance, 13(2-3), 392-420. Füss, R., Hille, J., Rindler, P., Schmidt, J. & Schmidt, M. (2010). From Rising stars and falling angels: on the relationship between the performance and ratings of German mutual funds. Journal of Wealth Management, 13(1), 75-90. Garnier, O., & Pujol, T. (2007). Can today's stars be used to read the stars of the future? Case study of equity funds distributed in France between 2000 and 2005, Working Papers Autorité des marchés financiers. Available at http://www.amffrance.org/technique/multimedia?docId=workspace://SpacesStore/b1fae1f7-a61f-4bb1-b7ee5f920d2a99ca_en_1.0_rendition. Gerrans, P. (2006). Morningstar ratings and future performance. Accounting & Finance, 46(4), 605-628. Goh, J., Ghosh, A. & Seng, W. (2012). Grades Matter in Performance: Morningstar Stewardship Grades and Mutual Fund Performance. Finance and Financial Manegement Commons. Available at: http://ink.library.smu.edu.sg/lkcsb_research/3324 Goldreyer, E. & Diltz, D. (1999). The Performance of Socially Responsible Mutual Funds: Incorporating Sociopolitical Information in Portfolio Selection. Managerial Finance, 25(1), 23-36. Gottesmam, A. & Morey, M. (2012). Mutual Fund corporate culture and performance. Review of Financial Economics, 21(2), 69-81. Gregory, A., Matatko, J., & Luther, R. (1997). Ethical until trust financial performance: Small company effects and fund size effects. Journal of Business Finance and Accounting, 24(5), 705–725. Gregory A & Whittaker, J.M. (2007). Performance and performance persistence of ethical unit trusts in the UK. Journal of Business Finance and Accounting, 34(7-8), 1327–1344. Grinblatt, M., & Sheridan, T. (1992). The Persistence of Mutual Fund Performance. Journal of Finance, 47(5),1977–1984. Grinblatt, M., Titman, S. & Wermers, R. (1995). Momentum investment strategies, portfolio performance, and herding: A study of mutual fund behaviour. American Economic Review, 85(5), 1088–1105. Grinblatt, M. & Keloharju, M. (2000). The investment behavior and performance of various investor types: a study of Finland's unique data set. Journal of Financial Economics, 55(1), 43-67. Are the ratings useful tools selecting mutual funds? 122 Gruber, M. (1996). The growth in actively managed mutual funds. The Journal of Finance, 51 (3), 783–810. GSIA (2017): 2016 Global sustainable Investment Review, Available at https://www.msci.com/esg-fund-metrics Hamilton, S., Jo, H. & Statman, M. (1993). Doing Well While Doing Good? The Investment Performance of Socially Responsible Mutual Funds. Financial Analysts Journal, 49(6), 62-66. Haslem, J. A. (2014). Morningstar Mutual Fund Measures and Selection Model. Journal of Wealth Management, 17(2), 19-30. Hendricks, D., Patel, J. & Zeckhauser, R. (1993). Hot hands in Mutual Funds: Short-Run Persistence of Relative Performance. Journal of Finance, 48(1), 93-130. Hazenberg, Z.Z. (2012). Investment Fund Governance: An empirical investigation of Luxembourg UCITS. Thesis at University of Luxembourg. Available at http://orbilu.uni.lu/handle/10993/15562 Hill, R.P., Ainscough, T., Shank, T. & D. Manullang (2007). Corporate social responsibility and social responsible investing: A global perspective. Journal of Business Ethics, 70(2), 165–174. Howe, T., & Pope, R. (1996). Equity Mutual Fund Historical Performance Ratings. Journal of Financial And Strategic Decisions, 9(1), 33-43. Humphrey, J.E., Warren, G.J. & Boon, J. (2016). What is Different about Socially Responsible Funds? A Holdings-Based Analysis. Journal of Business Ethics, 138(2), 263-277. Junkus, J. & Berry, T. D. (2015). Socially responsible investing: a review of the critical issues. Managerial Finance, 41(11), 1176-1201. Jun, X., Li, M., Yan, W., & Zhang, R. (2014). Flow-Performance Relationship and Star Effect: New Evidence from Chinese Mutual Funds. Emerging Markets Finance and Trade, 50(3), 81-101. Kamal, R. (2013). Can Morningstar Analyst Ratings Predict Fund Performance?. The Journal of Applied Business Research, 29(6), 1665-1672. Kelly, B. & Jiang, H. (2014). Tail Risk and Asset Prices. The Review of Financial Studies, 27(10), 2841-2871. Khorana, A., Servaes, H. & Wedge, L. (2007). Portfolio manager ownership and fund performance. Journal of Financial Economics, 85(1), 179–204. Khorana, A., Tufano P. & Wedge, L. (2007). Board structure, merger and shareholder wealth.: a study of mutual fund industry. Journal of Financial Economics, 85(2), 571-598. Koenker, R., & Basset, G. (1978). Regression Quantiles. Econometrica, 46(1), 33-50. Apenddix 123 Kong, S. X. & Tang, D. Y. (2008). Unitary boards and mutual fund governance. Journal of Financial Research, 31(3), 193-224. Kreander, N., Gray, R.H., Power, D.M. & Sinclair, C.D. (2002). The Financial Performance of European Ethical Funds 1996-1998, Journal of Accounting and Finance, 1, 3-22. Kreander, N., Gray, R.H., Power, D.M. & Sinclair, C.D. (2005). Evaluating the Performance of Ethical and Non-ethical Funds: A Matched Pair Analysis. Journal of Business Finance and Accounting, 32(7-8), 1465-1493. Kryzanowski, L. & Mohebshahedin, M. (2016). Board governance, monetary interest, and closed – end fund performance. Journal of Corporate Finance, 38, 196-217. Kurtz, L. (1997). The Impact of Social Screening on Growth-Oriented Investment Strategies. The Journal of Performance Measurement, Spring, 65-71. Lai, S., Tiwari A. & Zhang, Z. (2010). Mutual Fund Flows, Performance Persistence, and Board Quality. Available at SSRN: https://ssrn.com/abstract=1570525 Lean, H.H., Ang, W.R. & Smyth, R. (2014). Performance persistence of socially responsible investment funds in the Asia Pacific region in G. Gregoriou and L. David (Eds.), Handbook of Asian finance REITs, trading, and fund performance, Academic Press Inc, Oxford, 377–392. Lean, H.H., Ang, W.R. & Smyth, R. (2015). Performance and performance persistence of socially responsible investment funds in Europe and North America. The North American Journal of Economics and Finance, 34, 254–266. Leite, P. & Cortez, M.C. (2014). Style and performance of international socially responsible funds in Europe. Research in International Business and Finance, 30(1), 248–267. Leite, P. & Cortez, M.C. (2015). Performance of European Socially Responsible Funds during Market Crises: Evidence from France. International Review of Financial Analysis, 40, 132-141 Lewis, A., & Mackenzie, C. (2000). Morals, Motives and Money: The Case of U.K. Ethical Investing. Human Relations, 53(2): 179–191. Luther, R.G., Matatko, J. & Corner, D.C. (1992). The Investment Performance of UK 'Ethical' Unit Trusts’. Accounting, Auditing and Accountability Journal, 5(4), 57-70. Mallin, C.A., Saadouni, B. & Briston, R.J. (1995). The financial performance of ethical investment funds. Journal of Business Finance and Accounting, 22(4), 483-496. Mamatzakis, E. & Xu, B. (2017). Does corporate governance matter in fund management company?: the case of China, Available at https://mpra.ub.unimuenchen.de/76138/1/MPRA_paper_76138.pdf Are the ratings useful tools selecting mutual funds? 124 Margolis, J.D., Elfenbein, H.A., & Walsh, J.P. (2009). Does it pay to be good? A metaanalysis and redirection of research on the relationship between corporate social and financial performance. Available at SSRN: https://ssrn.com/abstract=1866371. Meinhardt, C. (2014). Ratings and Performance of German Mutual Funds – A Comparison of Feri Trust, Finanztest and FondsNote. Available at SSRN: http://ssrn.com/abstract=1943577. Meschke, F. (2007). An empirical examination of mutual fund boards. Available at SSRN: http://ssrn.com/abstract=676901 Morey, M. (2002). Mutual Fund Age and Morningstar Ratings. Financial Analysts Journal, 58 (2), 56-63. Morey, M. (2005). The Kiss Of Death: A 5-Star Morningstar Mutual Fund Rating?. Journal of Investment Management, 3(2), 41-52. Morey, M., & Gottesman, A. (2006). Morningstar Mutual Fund Ratings Redux. Journal of Investment Consulting, 8(1), 25–37. Morningstar (2010). The Morningstar Stewardship Grade for Funds. Available at http://corporate.morningstar.com/us/documents/stewardshipgradefunds/StewGradeMethodology _06-18-07.pdf Morningstar (2011). Morningstar Analyst Rating for Funds Methodology Document. Available at https://corporate.morningstar.com/us/documents/MethodologyDocuments/AnalystRatingforFund sMethodology.pdf Morningstar (2016a). The Morningstar Sustainable Investing Handbook. Available at http://corporate1.morningstar.com/Morningstar-Sustainable-Investing-Handbook.pdf Morningstar (2016b). The Morningstar Sustainability Rating. Available at http://corporate1.morningstar.com/Morningstar-Sustainable-Rating-Methodology.pdf Müller, S., & Weber, M. (2014). Evaluation the Rating of Stiftung Warentest: How good are the Ratings and can they be improved?. European Financial Management, 20(2), 207-235. Muñoz, F., Vargas, M. & Marco, I. (2014). Environmental Mutual Funds: Financial Performance and Managerial Abilities. Journal of Business Ethics, 124(4), 551–569. Nanigian, D. (2012). Why do mutual funds expenses matter?. Financial Services Review, 21(3), 239 –257. Ng, W.S. (2009). Does Morningstar Shine in the Universe of Mutual Funds? A Study on Morningstar Mutual Fund Ratings. Singapore Management University Master Thesis, Available at: http://ink.library.smu.edu.sg/etd_coll/10 Apenddix 125 Nofsinger, J. & Sias, R. (1999). Herding and Feedback Trading by Institutional and Individual Investors. Journal of Finance, 54(6), 2263-2295. Nofsinger, J. & Varma, A. (2014). Socially responsible funds and market crises. Journal of Banking and Finance, 48, 180–193. Philips, C., & Kinniry, F. (2010). Mutual fund ratings and future performance. Vanguard, 1-12, Available at http://www.vanguard.com/pdf/icrwmf.pdf. Qian, M. (2006). Whom Can You Trust? A Study on Mutual Fund Governance. Available at SSRN: https://ssrn.com/abstract=685543 Renneboog, L., Ter Horst, J. & C. Zhang. (2008a). The price of ethics and stakeholder governance: The performance of socially responsible mutual funds. Journal of Corporate Finance, 14(3), 302–322. Renneboog, L., Ter Horst, J & C. Zhang (2008b). Socially responsible investments: Institutional aspects, performance, and investor behavior. Journal of Banking and Finance, 32(9), 1723–1742. Renneboog, L, Horst, J & Zhang, C. (2011). Is Ethical Money Financially Smart? Nonfinancial Attributes and Money Flows of Socially Responsible Investment Funds. Journal of Financial Intermediation, 20(4), 562-588. Rudd, A. (1981). Social Responsibility and Portfolio Performance. California Management Review, 23(4), 55–61. Sauer, D. (1997). The Impact of Social-Responsibility Screens on Investment Performance: Evidence from the Domini 400 Social Index and Domini Equity Fund, Review of Financial Economics, 6(2), 137-149. Scholtens, B. (2005). Style and performance of Dutch social responsible investment funds. Journal of Investing, 14(1), 63-72. Statman, M. & Glushkov, D. (2016). Classifying and Measuring the Performance of Socially Responsible Mutual Funds. Journal of Portfolio Management, 42(2), 140-151 Statman, M. (2000). Socially Responsible Mutual Funds. Financial Analysts Journal, 56(3), 30-39. Sustainalytics (2016). Sustainalytics’ Research Methodology. Company ESG Research. Trahan, E.A. (2008). Mutual fund governance and fund performance. Corporate Ownership & Control, 5(4),384-392. Tufano, P. & Sevick, M. (1997). Board structure and fee-setting in the US mutual fund industry. Journal of Financial Economics, 46(3), 321–355. Vidal-García, J. (2013). The persistence Of European Mutual Fund Performance. Research in International Business and Finance, 28, 45-67. Are the ratings useful tools selecting mutual funds? 126 Webley, P., Lewis, A. & Mackenzie, C. (2001). Commitment Among Ethical Investors: An Experimental Approach. Journal of Economic Psychology, 22(1), 27-42. Wellman, J. W. & Zhou, J. (2008). Corporate Governance and Mutual Fund Performance: A First Look at the Morningstar Stewardship Grades. Available at SSRN: https://ssrn.com/abstract=1107841 Wermers, R. (1999). Mutual Fund Herding and the Impact on Stock Prices, Journal of Finance, 54(2), 581-622. Wimmer, M. (2012). ESG-Persistence in Socially Responsible Mutual Funds. Available at SSRN: https://ssrn.com/abstract=2138882 White, M. (1995). The Performance of Environmental Mutual Funds in the United States and Germany: Is there Economic Hope for “Green” Investors?. Research in Corporate Social Performance and Policy, Supplement 1, 323344. Apenddix 127 APENDDIX Table 37Cross sectional Yearly Alpha Ratio Variable 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 l1.stars4 0.3465 10.546 -2.6092** 0.7886 -0.1326 -0.0639 -1.3805* -1.3743** -1.7699*** -1.6551*** -0.9953*** l1.stars3 -2.0737* 0.0908 -12.671 1.3468* -14.491 -0.6517 -1.7511** -1.7412*** -3.1438*** -3.1120*** -1.8293*** l1.stars2 -2.4073* 0.5101 -0.6558 1.8568** -1.9612* -0.3558 -1.7153** -2.3797*** -3.5973*** -4.4160*** -2.7856*** l1.stars1 -4.1337*** -0.1187 -3.4873*** -0.4293 -17.086 0.0717 -2.5637** -4.1585*** -4.9201*** -6.1315*** -4.3172*** Largecapblend 0.9609 1.4079* -15.676 -2.0834** 13.347 3.7379*** 2.8036*** -0.9174 0.4702 -0.4582 0.1917 Largecapgrwth 23.104 0.2374 18.409 -17.055 4.4238** 3.4828* 3.2692* -13.669 -0.2775 -0.605 -0.4909 Largecapvalue -0.8598 0.0889 -2.2598* 0.7169 4.2876*** 2.7899** 2.4619** -1.7742** 0.6739 -1.3897* 0.4641 Midcap 0.7193 0.3748 -2.8022* -5.5996*** 23.929 4.5212*** 3.0099* -16.036 15.813 0.0789 0.1248 Smallcap 15.652 2.6694** -0.6278 1.13 5.6740*** 3.7871** 4.1412*** -2.5002*** 1.5769* 0.2854 0.2347 LargecapexUK 10.493 1.7596* -0.5128 -0.3624 2.7413* 2.8660** 2.9389*** -1.5364** 0.461 -12.437 0.6278 SmallmidexUK -93.953 -1.356 1.603 -2.6657*** 8.8624* 0.998 29.355 -3.8539*** 0.6825 -21.461 0.486 Eurozoneflexcap -82.692 11.421 -49.953 -0.5315 0.7524 4.3149** 2.178 0.8248 0.3369 0.3386 1.2384* Eurozonelargecap 0.7913 0.8467 -13.781 -1.5495** 2.5217* 3.9301*** 3.1893*** -1.0458* 0.9951 -0.0717 0.4496 Eurozonemidcap 1.427 0.6855 -28.821 -11.041 22.992 3.202 20.259 0.4654 -13.045 0.3059 2.8614*** Eurozonesmallcap -14.2839** 7.1779** -10.1311** -43.192 -12.504 20.901 42.101 -0.8958 0.8402 11.521 -0.6671 _cons -0.5581 -1.6639* 24.779 0.9535 -0.9151 -3.2951** -16.555 3.0024*** 1.7408** 2.7275*** 1.0554** N 352 405 473 514 667 778 911 1029 1145 1241 1345 r2 0.1988 0.0672 0.105 0.0675 0.0635 0.0407 0.0324 0.0624 0.0748 0.1242 0.0956 F 36.165 19.334 35.968 32.748 24.655 14.336 1.364 35.124 46.368 97.084 75.243 This table reports the coefficients for Cross sectional Yearly models for Alpha L1.Star is the one year lagged variable representing the rating of the mutual fund and yr* are the time dummies variables and, finally, Largecapblend, Largecapgrwth, Largecapvalue, Midcap, Smallcap, LargecapexUK, SmallmidexUK, Eurozoneflexcap, Eurozonelargecap, Eurozonemidcap and Eurozonesmallcap are dummies to control for categories. N is the number of observations, r2 the pseudo-squared fit measure, Rho is the fraction of variance due to individual effects and Chi-square (p) is the p-value associated to the Chi-square significance test. *Significant at 10%; ** significant at 5% and *** significant at 1%. Are the ratings useful tools selecting mutual funds? 128 Table 38Cross sectional Yearly Sharpe Ratio Variable 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 stars4 -0.1610* -0.2085*** -0.3038*** -0.0531 -0.0461** -0.0829*** -0.1108*** -0.1725*** -0.2141*** -0.1420*** -0.2309*** stars3 -0.3343*** -0.4242*** -0.3818*** -0.1388*** -0.1184*** -0.1323*** -0.1784*** -0.2538*** -0.3597*** -0.2875*** -0.4019*** stars2 -0.4028*** -0.5052*** -0.5532*** -0.1619*** -0.1843*** -0.1854*** -0.2383*** -0.3071*** -0.4625*** -0.4254*** -0.6129*** stars1 -0.4361*** -0.5878*** -0.9026*** -0.4209*** -0.2862*** -0.2644*** -0.3137*** -0.4251*** -0.6081*** -0.5774*** -0.8425*** Largecapblend -0.2881** -0.2376** -0.0718 0.0599 -0.0777*** -0.0431 -0.1270*** -0.0614* -0.0473 0.0055 -0.0639 Largecapgrwth -0.4270*** -0.2553** 0.0111 0.2502*** 0.0418 0.0159 0.0023 0.1883*** 0.2163*** 0.1246** 0.0323 Largecapvalue -0.3281** -0.2873** -0.0895 0.0167 -0.0415 -0.0265 -0.1505*** -0.2046*** -0.1603*** -0.0866*** -0.2378*** Midcap 0.0301 0.4201*** 0.2372** -0.0014 0.0939*** 0.0145 0.1308*** 0.1991*** 0.1992*** 0.0596 0.1965*** Smallcap 0.1428 0.4887*** 0.2778** 0.0594 0.0494 -0.0053 0.1162*** 0.2886*** 0.2101*** 0.1208*** 0.2392*** LargecapexUK -0.21 -0.0869 0.0975 0.3015*** 0.1076*** 0.0423 -0.0881*** -0.1774*** -0.0522 0.0550* 0.1526*** SmallmidexUK 0.0064 -0.0036 0.241 0.1172 0.2622*** 0.1699*** 0.1955*** 0.0608 0.1791** -0.0399 0.2676*** Eurozoneflexcap -0.0729 -0.1664 -0.1458 -0.1561 0.024 0.0104 -0.1093** -0.3012*** -0.2605*** -0.1839*** -0.2345*** Eurozonelargecap -0.2091 -0.2826** -0.1794** 0.1366*** 0.0585** 0.0423 -0.2077*** -0.4000*** -0.3208*** -0.1399*** -0.2861*** Eurozonemidcap 0.1976 0.1366 -0.1167 0.0261 0.1661*** 0.0713 0.0212 -0.0573 -0.0424 -0.0636* 0.0096 Eurozonesmallcap -0.2733 -0.219 -0.4042*** -0.2577* 0.0166 0.0417 0.0582 0.1284* -0.0098 -0.0338 -0.089 _cons 0.1402 1.7416*** 2.1054*** 1.1975*** -0.7648*** -0.4515*** -0.1105*** 0.7418*** 0.6914*** 0.8164*** 1.8866*** N 347 399 466 507 659 769 902 1020 1137 1233 1337 r2 0.3619 0.5743 0.3939 0.1873 0.362 0.2712 0.4698 0.536 0.5668 0.4725 0.5191 F 17.7361 32.383 26.9785 8.4156 26.1943 18.1597 48.0609 76.9651 112.4071 74.7891 93.9336 This table reports the coefficients for Cross sectional Yearly models for Sharpe Ratio. L1.Star is the one year lagged variable representing the rating of the mutual fund and yr* are the time dummies variables and, finally, Largecapblend, Largecapgrwth, Largecapvalue, Midcap, Smallcap, LargecapexUK, SmallmidexUK, Eurozoneflexcap, Eurozonelargecap, Eurozonemidcap and Eurozonesmallcap are dummies to control for categories. N is the number of observations, r2 the pseudo-squared fit measure, Rho is the fraction of variance due to individual effects and Chi-square (p) is the p-value associated to the Chi-square significance test. *Significant at 10%; ** significant at 5% and *** significant at 1%. Apenddix 129 Table 39Cross sectional Yearly Return models Variable 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 l1.stars4 0.7335 1.6293 -2.4431** 0.8042 0.7319 2.2895** -1.3259 -2.5281*** -0.7471 0.388 -1.0090*** l1.stars3 -1.4163 0.4328 -3.3198*** 1.3370* 0.1859 2.8690*** -1.8637** -3.2992*** -1.8471*** 0.4583 -1.9942*** l1.stars2 -1.3114 1.1088 -4.8222*** 1.7364** 1.1811 3.6653*** -1.8536** -3.8339*** -2.3018*** -0.338 -2.9877*** l1.stars1 -2.3680* 0.4264 -7.1562*** -0.3202 2.7388** 4.5570*** -2.7818** -6.5673*** -3.4163*** -1.6670** -4.2693*** Largecapblend -4.7312*** -2.2261* -2.6686*** 1.0549 0.6192 -3.8326** -6.2773*** 3.5520*** -1.2996* -3.0145*** 0.3428 Largecapgrwth -5.5211*** -2.2545 -2.2082** 3.0533* 3.5769** -2.5892 1.4701 5.3799*** 2.1375* -5.0105*** 0.921 Largecapvalue -4.5560** -4.1675*** -2.2491* 1.0513 2.2630* -5.7626*** -9.8041*** 1.6714** -1.9416** -1.1895 -1.1311* Midcap 6.4307*** 4.0762** 7.9774*** -3.3401** -2.0352 7.8834*** 8.5784*** -1.9007* 6.0698*** 2.1037* 1.0117 Smallcap 6.8630*** 8.7397*** 9.5096*** -2.0752 -5.1122*** 11.6623*** 13.1273*** -2.8697*** 5.9355*** 8.0255*** 0.2104 LargecapexUK -1.8079 2.2764* 0.9584 3.9411*** 3.3522*** -4.5163** -6.0746*** 1.1895 4.1295*** 0.6479 1.8342*** SmallmidexUK -2.1378 3.9267 8.0899* -0.7853 2.2172 10.7761 11.8432*** -4.6959*** 4.0167* 4.0394 1.0661 Eurozoneflexcap -6.3226 -0.7626 -4.7742 0.1973 0.1172 -3.6653 -7.6905*** -4.9109*** -0.1656 0.3059 -1.8129** Eurozonelargecap -4.4046** -4.0633*** -2.9421*** 5.1678*** 2.7376*** -7.4670*** -14.6624*** -0.8968 -0.0237 -0.6841 -1.8558*** Eurozonemidcap 5.3695*** 2.6055 7.7454*** -0.4488 -0.2028 6.2361* 3.3313 -4.0323*** 2.5656 3.2664** -0.172 Eurozonesmallcap -4.4757 12.7615*** -4.6467** -3.5651 -3.1534 6.9809 4.7086 -3.4885** 0.165 4.3030*** -1.4365 _cons 14.2087*** 26.2073*** 23.6132*** -0.8332 -45.2903*** 31.2747*** 18.9595*** -11.6638*** 20.2903*** 22.3123*** 6.3030*** N 352 405 473 514 667 778 911 1029 1145 1241 1345 r2 0.421 0.3409 0.4123 0.1921 0.109 0.2695 0.5515 0.2883 0.1874 0.1969 0.1731 F 23.0922 13.6175 19.7274 12.6096 6.0314 18.3487 60.7363 24.1035 18.343 16.9468 21.2588 This table reports the coefficients for Cross sectional Yearly models for Yerly return. L1.Star is the one year lagged variable representing the rating of the mutual fund and yr* are the time dummies variables and, finally, Largecapblend, Largecapgrwth, Largecapvalue, Midcap, Smallcap, LargecapexUK, SmallmidexUK, Eurozoneflexcap, Eurozonelargecap, Eurozonemidcap and Eurozonesmallcap are dummies to control for categories. N is the number of observations, r2 the pseudo-squared fit measure, Rho is the fraction of variance due to individual effects and Chi-square (p) is the p-value associated to the Chi-square significance test. *Significant at 10%; ** significant at 5% and *** significant at 1%. .