Fund Scorecards Methodology Morningstar's Due Diligence Reports

Transcription

1 ? Methodology Morningstar's Due Diligence Reports Morningstar Due Diligence Reports 1 January 2017 Version 1.2; 1 October 2017 Introduction At Morningstar s core is research and data, but transparency and clarity explain why investors have trusted us for decades. - Kunal Kapoor, Morningstar CEO Contents 1 Introduction 2 Best Practices 4 Fund Scorecard Results 6 Fund Scorecard Factors 8 Scoring Factors Selection 14 Evaluating the Results 18 Appendix I: Morningstar's Ratings 19 Appendix II: Definitions 23 Disclosures Michael Laske Manager Research & Due Diligence Reports Product Manager michael.laske@morningstar.com For institutions, advisors, and investors, the Due Diligence Scorecards for funds provide the best forward-looking insights and transparency into the mutual fund and ETF universe, because they combine two proven approaches for evaluating funds into one easy-to-read report. Key Benefits Data-Driven Analysis for Comparing Funds With the ability to choose from more than 8,000 funds, investors face a paradox of choice. The Fund Scorecards solve this challenge by bringing together Morningstar's strengths research and data into a unique solution and help people calibrate their own views. They provide an efficient method for investors who want to use data-driven analysis as an input into their manager selection process. The Best of Both Worlds: Morningstar's Qualitative and Quantitative Evaluations From a qualitative perspective, Morningstar's manager research group takes a holistic look at the fund and focuses on five pillars, which our analysts believe identify the strategies most likely to outperform their peers over the long run on a risk-adjusted basis. Final decisions are based on interviews with the portfolio manager. From a quantitative perspective, the were built to succeed on three dimensions: (1) Efficacy, (2) Transparency, and (3) Broad Coverage. The scorecards aim to capture as much forecasting power as possible without losing interpretation and understanding. They are in no way substitutes for analyst judgment, but often provide valuable insights and serve as a very useful focal point in manager selection discussions. Analyst Rating Scorecard Ranking Source Manager Research Group Due Diligence Scorecards Approach Qualitative Quantitative Results Five Tiers Numerical Measure Absolute Relative Judgment Subjective Objective Framework Bottom-Up Top-Down Fund Coverage 18% 92%

2 Page 2 of 25 Best Practices The structure the unstructured and convert raw data into numerical scores for openend mutual funds and ETFs, relative to other funds in their peer group. The scores are determined by a 12-factor formula, which varies by asset class, and accounts for fees, manager experience, and past performance, among other factors. In short, the scorecards take a lot of data and condense it into a ranking that can be interpreted and easily understood. The Scorecards increase the likelihood of selecting the best funds, relative to many other measures, such as past performance, because the scores depend on the most important criteria. We have done the hard work on investors' behalf to help them focus on the best candidates for investment. Further, the numerical rankings help investors with a more granular comparison, relative to Morningstar's five-tiered rating systems. It is worth cautioning that the Scorecards try to identify broad patterns, and not blind investors to the substantial heterogeneity in the fund industry. A key reason for these differences is that funds differ along many dimensions, some of which cannot be captured through data-driven analysis, and we ask that you bear this caution in mind as you evaluate funds with the Scorecards. Scorecard Reports The Fund Scorecard report provides a complete view of Morningstar's best research all in one location, instead of requiring numerous data sources and tools. Each Morningstar rating (for example, Analyst Ratings, Quantitative Ratings, Star Ratings, Sustainability/ESG Ratings) is included when available. For funds with analyst coverage, the Scorecards often confirm the high quality of our Medalists, and even reveal some hidden gems that aren't covered by our research group. Scorecard reports are distributed on the fifth business day of each month in an Excel file for all eligible mutual funds and ETFs. The reports include embedded tools to filter the universe to a specific list. Custom reports for a select list of funds and ETFs can be delivered in Excel or PDF format upon request. The Scorecards have six styles. The scoring criteria are adjusted for each style. 1. Active Equity (For U.S., international, and sector equity mutual funds) 2. Active Bond (For taxable and municipal bond mutual funds) 3. Passive (For equity and bond mutual funds) 4. Allocation (For active/passive target-date and balanced mutual funds) 5. Alternative (For active/passive liquid alternative mutual funds) 6. ETFs (For active/passive exchange-traded funds)

3 Page 3 of 25 Scorecard Templates Scorecard templates are accessible in Morningstar Direct for clients who license Morningstar's Due Diligence Reports and Morningstar Direct. Similar to the reports, there are six styles of templates. In Morningstar Direct, clients will need to first separate their full list of mutual funds and ETFs based on the six styles. Requests received by 8:00am each Monday will be processed by 5:00pm. The key difference between the reports and templates is the display of the scoring results. In the reports, the Scores, Ranks, and Quartiles are calculated for all funds, relative to each fund's peer group. When using a template, the Scores, Ranks, and Quartiles are calculated for the funds in an investor's customized list. When the reports and templates are used together, they can provide different perspectives. Note: Using the Scorecard Templates requires the Morningstar Direct users to divide their full list of funds by Scorecard style. This is done automatically when using the Scorecard reports. Peer Groups The Morningstar Category classification is selected as the default peer group. This organizes funds with similar objectives based on their portfolio statistics and compositions during the past three years. Benchmark Selection The Morningstar Category Index (for active funds) and the Prospectus Benchmark (for passive funds and ETFs) are selected as the default benchmarks. Fund Eligibility The Scorecard reports are intended to objectively compare all mutual funds and ETFs available to investors, with a few exceptions. The Scorecards include only funds with at least 50% representation of their respective scoring factors, including fees, at least one year of performance, and placement in a peer group with three or more funds. 1 Consistent with Morningstar's research methodology, funds placed in categories that are undefined or which have an objective that does not aim to provide high risk-adjusted returns are also excluded from the Scorecards. 2 1 Additional criteria are included for the different Scorecard styles. Active Equity and Active Bond funds must include data for the longest-tenured manager. Passive Equity/Bond, Allocation and ETFs must included data for portfolio turnover. Finally, Alternative funds must be assigned a Morningstar 'Star' Rating, which requires three years of performance data. 2 As of October 1, 2017, Morningstar classifies open-end mutual funds and exchange-traded funds into 123 categories. The exclude funds from the following categories: Money Market (4); Commodities (4); Bear Market, Multicurrency, Single Currency, Trading (7), and Volatility in the Alternative category group; Miscellaneous Region and Miscellaneous Sector in the Equity category group.

4 Page 4 of 25 Fund Scorecard Results Since we have never been able to identify a single variable that fund performance hinges on, we solve the problem by combining the most predictive measures, each with minimal predictive power, and extrapolate a total score. The whole is greater than the sum of its parts, and together, the Scorecards paint a more meaningful picture before selecting a manager. Score (Result #1) The total score is the result of aggregating the 12 scoring factors using a weighted average calculation. The scores range from 1 to 100. Each fund's score is a relative measure against all other funds in the peer group. Consistent with Morningstar's research methodology, a lower number is better. Funds with low scores are more likely to outperform similar funds over the long term. Conversely, funds with high scores are more likely to underperform similar funds over the long term. This is demonstrated later in the Evaluating The Results section. Rank (Result #2) Next to the Score, we rank each fund within its peer group. The fund with the best Score receives a ranking of 1. The fund with the second-best score receives a ranking of 2, and so on. Quartile (Result #3) Next to the Rank, we divide the peer group into quartiles to quickly identify funds of similar quality. Fund Scorecard Factors The scoring factors are on full display next to the results. Transparency is important to Morningstar and the scores are not a black box calculation. If a fund receives a high or low score, you will immediately know why. The factors are color-coded to highlight each fund's strengths and weaknesses. If a scoring factor is not included, this means the data point is unavailable. This will occur most often for funds which have a short track record. However, the performance factors receive relatively small weightings and the remaining factors are rescaled to calculate the Score. This step-by-step guide describes how the data for each scoring factor are used to determine a fund's total score. Calculation Guide Stability of the Scores The Scorecards are calculated monthly and we have observed stable results since their inception. The average monthly standard deviation for the Total Scores is 2.4 which reflects their consistency. As the performance measures in the Scorecards are calculated over long time periods, a significant change in the Total Score is often attributed to a significant change for the fund.

5 Page 5 of 25 Fee Change: An increase or decrease in the expense ratio can have a material impact on the score. Manager Change: The departure of a long-tenured portfolio manager can have a material impact on the score. Distribution of Scores Historically, the scores across all asset classes follow a normal distribution. Exhibit 1 Distribution of Scores (0-100) Source: Morningstar Research Services LLC. Scores calculated for 15,147 share classes. Data as of dates: 1/1/2017-6/30/2017.

6 Page 6 of 25 Fund Scorecard Factors The six Scorecard styles are designed to compare funds across different asset classes. The Total Scores are always determined by 12 scoring factors. Across the six Scorecard styles, a total of 28 scoring factors are used, and 12 are proprietary to Morningstar. The 12 proprietary scoring factors include Morningstar Risk-Adjusted Return, Morningstar Risk, Morningstar Fee Level, and Morningstar's firm-level calculations, among others. Each scoring factor represents one of the five pillars behind Morningstar's manager research framework. Exhibit 2 Factors Table Source: Morningstar Research Services LLC. The Equity and Bond Scorecards evaluate actively managed mutual funds.

7 Page 7 of 25 Factor Weights We take a different approach from many scoring measures in the fund research industry today because we place a relatively small importance on past performance. The weightings assigned to each factor are adjusted based on the factor's importance and were determined by methods described in the Factor Selection section of this document. Fee Factors When comparing funds across share classes, it is important to account for different fee structures. For example, when comparing a fund's A share class with its Institutional share class, different fee measures can have a material impact on a fund's Total Score. As a result, we provide the option to choose the Net Expense Ratio (default), the Morningstar Fee Level, or the Management Fee as the scoring factor for Fees. Performance Factors For scoring factors in the Performance pillar, the Scorecard features an Extended Performance setting, which leverages the historical performance of a fund if a share class does not have the required history. Missing Factors The scoring factors are standardized based on the percentile rank within a peer group. If Morningstar's database does collect data for a fund on one or more scoring factors, the null values are removed from the Total Score calculation and the remaining values are re-weighted to a total of 100%.

8 Page 8 of 25 Scoring Factors Selection The Scorecards are designed to compare funds on the most important factors. Today, Morningstar's flows data suggests that many investors select funds for reasons that don't correlate with future success, such as past performance. This contradicts what educated investors already know: Selecting funds based on past performance is like driving a car while looking only in the rear-view mirror. We believe that helping investors look forward with a richer toolkit will lead to greater investor success. The Scorecards were created by combining two frameworks. First, we leveraged the foundation of our manager research group's five-pillar approach and second, we performed data-driven analysis. Framework #1: Five-Pillar Approach When our research analysts evaluate a fund, their analysis of five pillars Price, Performance, Process, People, and Parent determines the rating. As a general starting point for their analysis, the People, Process, and Parent pillars are worth 25%, Performance is worth 15%, and Price is worth 10%. 1. Price Our research shows that fees are one of the most important measures for predicting future performance. In some asset classes, cost differences may account for up to 80% of the differences in returns. Investors must recognize the impact that excessive fees have on long-term investment results. Our research: The Predictive Power of Fees 2. Performance Past performance, specifically over the short term, is not predictive of future performance on average, and performance measures receive a relatively small weighting when we evaluate funds. This is supported by many academic studies which show that performance persistence is difficult to detect and, when detected, tends to be driven by persistent under-performance rather than over-performance. (Carhart et al., 2002). Further, our research shows that 1-star funds underperform their peers. Our research: The Predictive Power of the Star Rating 3. Process Our research shows that equity strategies with high turnover underperform over long time horizons on average. High turnover means high transaction costs and can also indicate that a portfolio manager has a short time horizon. We believe in a long-term investing philosophy. "Inactivity strikes us as intelligent behavior. Neither we nor most business managers would dream of feverishly trading highly-profitable subsidiaries because a small move in the Federal Reserve s discount rate was predicted or because some Wall Street pundit had reversed his view on the market. Why, then, should we behave differently with our [investments]?" Warren Buffett

9 Page 9 of People Managers who have run a strategy for many years often have more skill at implementing and perfecting their methods in different market environments. However, portfolio managers currently have an average tenure of only five years. Our research shows that funds run by experienced managers are less likely to encounter abrupt manager turnover, thereby mitigating manager risk. Our research: Manager Tenure: There's No Substitute for Experience 5. Parent Fund providers that put the needs of investors first, over and above the interests of any third party, typically succeed over the long term. These firms honor their commitment to serve as honest stewards of the assets of their investors. Those that retain their talented managers, rarely liquidate or merge funds, and keep costs low often have a strong culture of stewardship that rewards investors. Our research: Fund Family Research Framework #2: Data-Driven Analysis With thousands of data points to choose from, we identified a group of data points to represent each pillar, and constructed a dataset that included all U.S. mutual funds and ETFs that held investors assets between 2002 and The dataset contained every share class during the 14-year period, including those liquidated or merged, which ensured our dataset was free of survivorship bias. In addition to the quantitative variables, we included classification measures to differentiate funds by asset class, style, and active/passive. For historical risk/reward measures, we included data for the three-, five-, and 10- year trailing periods. To forecast risk-adjusted performance, we considered a 10-year prediction variable, but this would have reduced the number of funds in our dataset by nearly half. Unfortunately, many funds don t live for 10 years. Instead, we decided to use the five-year Sharpe ratio. We selected this measure because we believe the quality of a fund can be thought of as the expectation that it will outperform on a riskadjusted basis in the future. By selecting a risk-adjusted measure rather than a total return measure, we aimed to reward funds that achieved high performance without taking excessive risks. The explanatory variables included risk measures, performance measures, proprietary firm-level measures, and operational information. Funds that don't have a long history are often a challenge to evaluate, but if we know that a fund comes from a successful fund family, there is a higher likelihood that it will outperform in the future. Definitions for each variable are provided in the Appendix.

10 Page 10 of 25 Exhibit 3 Explanatory Variables Source: Morningstar Research Services LLC After our dataset was constructed, we standardized the values across each factor, with respect to a fund's category. For example, a large-cap value fund was compared only with other large-cap value funds. If a fund s standardized expense ratio had a value of 25, this meant the fund s expense ratio was in the 25th percentile among all funds in the same category. Our dataset had robust coverage for each variable and spanned more than 220,000 observations across 14 years. In the instances of missing data, we substituted the mean value for the category. To create the prediction variable, we shifted the time period backward five years for each five-year Sharpe ratio. This aligned the future, unknown Sharpe ratio with the explanatory variables from which we would make predictions. The reference periods in our analysis were rolling one-year intervals such that the explanatory variables would be measured ex ante, and the evaluation period was always five years later, measured ex post. We divided the observations into training, validation, and test sets to allow for out-of-sample testing. Exhibit 4 Time Periods Analyzed Source: Morningstar Research Services LLC

11 Page 11 of 25 Variable Correlation Testing To begin our analysis, we hypothesized that few variables would correlate with future performance in isolation. This was confirmed from a correlation plot that showed the prediction variable's correlation with the explanatory variables. A darker shade of red (positive) or blue (negative) reflects a stronger correlation. The far-right column reflects the prediction variable and displays a light shade for each variable, suggesting that these measures do not correlate with future performance on their own. Exhibit 5 Variable Correlations Source: Morningstar Research Services LLC. Data as of dates: 1/1/ /31/2015. Calculations use the Pearson method. Variable Importance Testing Once we confirmed that none of the factors showed strong correlation with future performance in isolation, we hypothesized that combining the measures with the highest correlation could extrapolate an aggregated score that was more predictive than any measure on its own. To identify the most important factors, we analyzed variable importance charts from a linear regression model. The results reflected two important insights. First, the expense ratio contributed the most to future performance. This is consistent with the theme in the current market environment that costs cannot be ignored when selecting funds. Second, none of the most important variables represented a fund's performance, which is often the one measure used by investors to compare funds.

12 Page 12 of 25 Exhibit 6 Variable Importance: Linear Model Source: Morningstar Research Services LLC. Includes variables with a t-value greater than 2. Data as of dates: 1/1/ /31/2015. We also analyzed results from a nonlinear random forest model, which were generally consistent with the linear model. This gave us greater conviction that we identified the most important factors. Exhibit 7 Variable Importance: Nonlinear Model Source: Morningstar Research Services LLC. Data as of dates: 1/1/ /31/2015.

13 Page 13 of 25 In both models, the least important variables were 10-year measures, which we believe is a reflection that 40%-50% of funds, on average, are liquidated or merged before they reach their tenth year. As nearly half of the funds didn't have 10 years of performance history, this eliminated much of these variables' importance. Building the Scorecards With the insights described above, we created the Scorecards by calibrating the art of our manager research framework with the science of quantitative testing. We were conscious of data mining and the danger of false precision, so the results from our quantitative testing influenced the selection of factors and weightings, but the final decisions were made independently and based on the guiding principles of our research framework.

14 Page 14 of 25 Evaluating the Results Using out-of-sample testing, we compared the Scorecards' results with each fund's future 5-Year Sharpe ratio, which we standardized over all time periods to allow for accurate comparisons. Exhibit 8 The Scorecards Correlate With Risk-Adjusted Performance Source: Morningstar Research Services LLC. Data as of dates: 1/1/ /31/2015. Includes active and passive equity, bond, & allocation funds. Exhibit 9 The Top-Scoring Funds Correlate With Better Risk-Adjusted Performance Source: Morningstar Research Services LLC. Data as of dates: 1/1/ /31/2015. Includes active and passive equity, bond, & allocation funds.

15 Page 15 of 25 In our experience, some investors are interested only in the best funds to own, so we evaluated funds that scored in the top 30% of their categories. During the 14-year period, there was a 78% probability that a top-scoring fund would perform better than its peer group average. Further, there was only a 6% probability that a top-scoring fund would perform in the bottom quartile. Comparison With Morningstar's Star Ratings Today, many investors select funds based on historical performance, evidenced by positive flows for 5- star funds. To compare the Scorecards with Morningstar Risk-Adjusted Returns, the data behind the Morningstar Rating for funds, we analyzed the success ratios for both measures. Success ratios capture the percentage of funds that survive and outperform the category average. The results suggest the Scorecards are more effective at predicting the best and worst funds in advance. Exhibit 10 The Scorecards Are Better at Identifying the Best Performers and Avoiding the Worst Performers Source: Morningstar Research Services LLC. Data as of dates: 1/1/ /31/2015. Calculations are provided for the Scorecards that were included in the initial release on 1/1/2017. The calculations behind this chart are provided here. The time horizon in our analysis included a bull market as well as the financial crisis that occurred from 2007 to Our results didn t show any significant changes before or after the financial crisis, nor did we find any indication that the importance of any variable changed significantly over time. This suggests the most important factors are as predictive today as they were in the 2000s and that they will continue to be important in the future.

16 Page 16 of 25 Correlation With Morningstar's Analyst Ratings The Scorecards codify Morningstar's manager research framework, and as a result, there is high correlation and a monotonic relationship with Morningstar's Analyst Ratings. Exhibit 11 Average Scores by Morningstar Analyst Rating Source: Morningstar Research Services LLC. Data as of 12/31/2016. The average scores are displayed for funds in the Scorecards that were included in the initial release on 1/1/2017. Future Testing While the Scorecards use a transparent, linear calculation, there are also nonlinear techniques that could be implemented to compare funds. We decided to evaluate our scoring framework against other methods. First, we established a baseline regression model that included every variable from the dataset and generated a root mean squared error, or RMSE, of We also tested a series of lasso regression models with cross-validation and received similar results. Next, we tested a series of random forest models and varied the robustness to include 500, 1,000, and 1,500 trees. We also implemented a boosting technique to capture more predictability from the most important variables. The smallest out-of-sample RMSE that we generated was We also explored neural network models that could potentially capture complex signals from the factors that were undetected in the other models. We tested different neural network models by tuning the parameters (for example, architecture and number of hidden layers, neurons, and epochs). The best neural network model generated an out-of-sample RMSE of 21.9.

17 Page 17 of 25 Exhibit 12 Error Rates (RMSE) for Different Models Source: Morningstar Research Services LLC. Data as of dates: 1/1/ /31/2015. Compared with the linear model, there could be benefits to using nonlinear models to increase predictive power. The Morningstar Quantitative Rating is an example of an assessment that uses a nonlinear model to rate mutual funds. However, it is worth noting the benefits of additional predictive power from a nonlinear model come at the cost of transparency and clarity. We will continue to evaluate the merits of these models in the future. Conclusion While it is impossible to predict every fund's future with absolute certainty, as a portfolio manager's decisions will shape a fund's future performance, the Scorecards succeed at ranking funds in an order that correlates with future risk-adjusted performance on average. K

18 Page 18 of 25 Appendix I: Morningstar's Ratings Morningstar's forward-looking ratings are provided, when available, in the. Œ This rating is assigned to funds with competitive advantages across all five pillars. Our analysts have the highest level of conviction that these funds will outperform their benchmark or peers, within the level of risk taken, over a full market cycle. This rating is assigned to funds with competitive advantages across most of the five pillars. Our analysts have a high level of conviction that these funds will outperform their benchmark or peers, within the level of risk taken, over a full market cycle. ª This rating is assigned to funds with competitive advantages across most of the five pillars. Our analysts have a sufficient level of conviction that these funds will outperform their benchmark or peers, within the level of risk taken, over a full market cycle. This rating is assigned to funds without any competitive advantages across the five pillars. Our analysts do not believe these funds will outperform or underperform their benchmark or peers, within the level of risk taken, over a full market cycle. Á This rating is assigned to funds with competitive disadvantages across one or more of the five pillars. Our analysts believe these funds will underperform their benchmark or peers, within the level of risk taken, over a full market cycle. ˆ This designation signals that a material change has occurred and the event requires further review to determine the impact on the rating. Our analysts aim to assign a new rating, or maintain the previous rating, within two weeks of the event. Methodology: Morningstar Analyst Rating for Funds Methodology: Morningstar Analyst Rating for ETFs Methodology: Morningstar Quantitative Rating Methodology: Morningstar 'Star' Rating

19 Page 19 of 25 Appendix II: Definitions Alpha (Performance) This factor measures the value added or subtracted by a fund's management team. It represents the difference between a fund's actual returns and its expected performance, given its level of risk as measured by beta. Beta (Performance) This factor measures the systematic risk that is based on the covariance of a fund's return with the return of the benchmark. A low beta does not imply that a fund has a low level of volatility, but it does suggest that a fund s index-related risk is low. Beta Standard Error (Performance) This factor measures how close a fund tracks an index. It calculates the volatility of a fund's returns after accounting for the volatility of the index. Calmar Ratio (Performance) This factor measures a fund's compound annualized rate of return during a specified time period divided by the absolute value of maximum drawdown during the same time period. Correlation (Performance) This factor measures the co-movement between a fund and the benchmark. Estimated Holding Cost (Price) This factor measures the realized performance of an ETF manager relative to the benchmark index after all expenses both disclosed and undisclosed. It represents the returns of the portfolio's net asset value relative to the underlying benchmark. A higher cost can contribute to greater underperformance. A smaller or even negative cost shows that the manager is doing a better job finding the lowest cost ways to replicate the benchmark index. This factor is a proprietary calculation for ETFs and requires one year of data. Expense Ratio (Price) The Prospectus Net Expense Ratio measures the annual percentage of a fund's assets paid out in expenses. Expenses include the management fee, 12b-1 fee, and all other asset-based fees associated with a fund's operations and distribution. In contrast to the Gross Expense Ratio, the Net Expense Ratio includes any voluntary or contractual waivers, reductions from brokerage service arrangements, and other expense reimbursements, expense offsets, or any arrangements of temporary nature.

20 Page 20 of 25 Fee Level (Price) The Morningstar Fee Level data are proprietary calculations of a share class' expense ratio relative to similar funds in the same category, distribution channel, and with a similar expense structure. Firm Average Fee Level (Parent) This factor measures the firm's average fee charged to investors. Firm Average Manager Tenure (Parent) This factor measures the average number of years that the portfolio manager of each fund has stayed with the firm. For funds with more than one manager, the tenure of the manager who has been with the fund the longest is included. Firm Manager Investment (Parent) This factor measures the percentage of fund assets where at least one portfolio manager has invested $1 million or more, as reported to the SEC. Firm Manager Retention (Parent) This factor measures the percentage of portfolio managers that stayed with the firm during the trailing five years. A new manager doesn t count against the rate, only departures. Our manager research group believes that a high rate reflects manager stability and a culture where few managers want to leave. Firm Risk-Adjusted Success Ratio (Parent) This factor measures the percentage of a firm's funds that were active five years ago, have not been liquidated or merged, and delivered a Morningstar Risk-Adjusted Return better than the median fund in their respective categories during the trailing five years. This ratio not only captures performance but also avoids survivorship bias; fund families are penalized for liquidating or merging their poorestperforming funds or those that didn t gain enough traction with investors to remain viable. Although in some cases, these decisions may be the right ones, our manager research group believes that a firm which regularly sweeps weak performance records under the rug or habitually launches and then merges away unsuccessful funds (i.e., fund churn) may not have investors best interests at heart. Information Ratio (Performance) This factor measures how much a fund outperformed its benchmark per level of risk. It is helpful for evaluating actively managed strategies that deviate from their benchmark to add alpha. Management Fee (Price) The management fee is the maximum percentage deducted from a fund's average net assets to pay an advisor or subadvisor.

21 Page 21 of 25 Manager Tenure (Longest) (People) This factor measures the number of years that the longest-tenured portfolio manager has been managing the fund. Manager Tenure (Average) (People) This factor measures the average number of years that the current portfolio management team has been managing the fund. Market Impact Cost (Price) This factor measures the liquidity of an ETF. Less liquid securities are more thinly traded and a single large trade can move their prices considerably. This calculation is an estimate of the basis point change in an ETF s price caused by a $100,000 trade. A lower market impact implies the ETF is more liquid, but the actual size of price movements due to a single trade may vary considerably from this estimate depending on market activity. The market impact cost for an ETF incorporates the information of the bidask spread and the depth of market order books by measuring the volatility of market prices around the true portfolio value. This calculation standardizes the volatility to a basis point value from a $100,000 trade through the common assumption that trades in the ETF are sequentially independent and return variance caused by the trade increases in linear proportion with the dollar size of each trade. This factor is a proprietary calculation for ETFs and requires one year of data. Max Drawdown (Performance) This factor measures a fund s maximum loss (%) in a peak-to-trough decline before a new peak is attained. It represents the downside risk during a specified time period. Morningstar Risk (Performance) This factor measures a fund s downside volatility against its peers in the same Morningstar Category. Morningstar Risk-Adjusted Return (Performance) This factor measures how a fund has performed on a risk-adjusted basis against its peers in the same Morningstar Category. Omega Ratio (Performance) This factor measures a fund's risk-adjusted return, similar to the Sharpe ratio. In contrast to the Sharpe ratio, it does not assume a normal return distribution. Portfolio Concentration (Process) This factor measures the idiosyncratic non-market risk taken on by an ETF. Since higher idiosyncratic risks are not always compensated with higher returns, a lower portfolio concentration is better for investors seeking to avoid security-specific and sector-specific risks. This measure is calculated for U.S. equity, international equity, and fixed-income ETFs, as those broad asset classes have diversified return

22 Page 22 of 25 factors generally agreed upon in the academic finance literature. Equity ETFs use proxies for U.S. or global market, size, and value risk factors to separate diversified, systemic portfolio risk from idiosyncratic risk. Fixed-income ETFs use proxies for market, credit, and duration risk factors to separate diversified, systemic portfolio risk from idiosyncratic risk. This factor is a proprietary calculation for ETFs and requires one year of data. R-Squared (Performance) This factor measures the correlation between the returns of a fund and its benchmark. Sharpe Ratio (Performance) This factor measures a fund's risk-adjusted return and signals how much volatility a fund had to take to earn excess return over the risk-free rate. It is calculated by dividing a fund's excess return by the standard deviation. Sortino Ratio (Performance) This factor measures a fund's risk-adjusted return, similar to the Sharpe ratio. In contrast to the Sharpe ratio, it differentiates downside volatility from total volatility by using downside deviation as the risk measure. Tracking Error (Performance) This factor measures the volatility of the differences between a fund's returns and the benchmark's returns. While most passive funds attempt to replicate the performance of an index, the divergence is captured by the tracking error. Tracking Volatility (Performance) This factor measures the random variation in an ETF against its benchmark. This seeks to isolate the ability of a portfolio manager to track the index, ignoring the confounding effects of liquidity on market prices. The calculation compares NAV returns, rather than market returns, to the benchmark. The statistical model includes a lagged error term to account for stale prices that can cause spurious daily pricing differences against many fixed-income and foreign equity indices. Lower tracking volatility shows better replication of the benchmark by the manager. This factor is a proprietary calculation for ETFs and requires one year of data. Turnover Ratio (Process) This factor measures portfolio turnover and represents a portfolio manager's trading activity. It is calculated by taking the lesser of purchases or sales, excluding all securities with maturities of less than one year, and dividing by the average monthly net assets. It is often used to estimate the percentage of a fund's holdings that have changed during the trailing year. Turnover data are sourced from a fund's annual report.

23 Page 23 of 25 Disclosures Timeline Version 1.1: January 1, 2017 Initial release of the for Active Equity, Active Bond, Passive, and Allocation funds. The Fund Scorecard for ETFs was released on April 1, Version 1.2: July 1, 2017 Minor weight adjustments applied to the performance-based scoring factors. Initial release of the Fund Scorecard for Alternatives funds. Future Versions Morningstar will adjust the scoring methodology infrequently, but potential improvements will be evaluated on an ongoing basis. Any modifications will likely be attributed to evolving market conditions or new insights from Morningstar's manager research group. We will provide notification in advance of future changes. Building on the foundation of the Scorecards for U.S. mutual funds and ETFs, the European Manager Research group modified the ' factors to function in European markets. European Scorecards are currently in development. Acknowledgments I would like to express my appreciation to Russel Kinnel, director of Morningstar's Manager Research Ratings Committee, for sharing his time with me to brainstorm on the most important factors worth consideration. Our Mission The align with Morningstar's mission to create the most effective investment data, research, and ratings for investors. Consistent with Morningstar's guiding principles, the Fund Scorecards put investors first, reflect our independent research framework, focus on the long term, reward low costs, and help investors build portfolios holistically. References Carhart, M "On Persistence in Mutual Fund Performance." The Journal of Finance, Vol LII, Cohen, R., Coval, J., & Pastor, L Judging Fund Managers by the Company They Keep.

24 Page 24 of 25 Kinnel, R Predictive Power of Fees: Why Mutual Fund Fees Are So Important. Morningstar Research Services LLC Morningstar Medalists: What Makes the Best Funds Shine. Shine.html Ptak, J Winning Funds Often Look Like Losers. Ptak, J Do Active Funds Belong in the 'Too Hard' Pile? Vanguard Research Shopping for alpha: You get what you don t pay for. Disclosures The qualitative and quantitative perspectives shared in the are intended to provide institutions, advisors, and investors with two approaches that complement each other and help individuals calibrate their own views based on their specific circumstances. Morningstar's manager research ratings are assigned by Morningstar Research Services LLC. Morningstar's forward-looking ratings are based on current expectations about future events and therefore involve unknown risks and uncertainties that may cause such expectations not to occur or to differ significantly from what was expected. The research analysts who contribute to Morningstar's forward-looking ratings abide by Morningstar's Code of Ethics. The Due Diligence Scorecards are intended to assist with the manager selection process, particularly when funds do not receive qualitative coverage from Morningstar Research Services LLC. The Due Diligence Scorecards do not provide an official rating from Morningstar Research Services LLC, and are intended for informational purposes only. Access to or use of the information contained in the Fund Scorecards does not establish an advisory or fiduciary relationship with Morningstar Research Services LLC, Morningstar Inc., or their content providers.

25 Page 25 of 25 About Morningstar Manager Research Morningstar provides independent, fundamental analysis on managed investment strategies. Analyst views are expressed in the form of Analyst Ratings, which are derived through research of five key pillars Process, Performance, Parent, People, and Price. A global research group issues detailed analyst reports on strategies that span vehicle, asset class, and geography. About Due Diligence Reports Morningstar's Due Diligence Reports are provided to institutions, advisors, and investors and serve as a toolkit to determine, demonstrate, and document a proven process for manager selection. Due Diligence Scorecards use data-driven analysis to compare mutual funds and ETFs against their respective peer groups. Due Diligence Watchlists monitor funds with red flags, and Alerts perform ongoing oversight. Considered a complement to Morningstar's software products, the Due Diligence Reports strengthen Morningstar's moat by incorporating proprietary data, research, a peer group classification system, and a proven scoring framework into one solution a combination unmatched in the manager research industry. For More Information Michael Laske Product Manager, Manager Research & Due Diligence Reports michael.laske@morningstar.com? 22 West Washington Street Chicago, IL USA