S E L L I N M A Y... A N D P A Y!

1945 1948 1951 1954 1957 1960 1963 1966 1969 1972 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 2011 Growth of $100 S E L L I N M A Y... A N D P A Y! SUMMARY It is that time of year. Articles in the financial and popular press are highlighting the initially compelling story of Sell in May and Go Away, also known as the Halloween Indicator. This strategy is based on a seasonal pattern of stock market returns being higher in the November through April timeframe than from May through October. A study by Bouman and Jacobsen ( The Halloween Indicator, Sell in May and Go Away: Another Puzzle ) shows that this is a long-term global phenomenon, and cites evidence of this trend in the United Kingdom market all the way back to 1694. So the question is: Should investors really Sell in May and Go Away? April 30, 2012 Northern Trust Global Investments James D. McDonald Chief Investment Strategist Daniel J. Phillips, CFA Investment Strategist Phillip B. Grant Investment Analyst Our research conclusively shows this to be a losing strategy. Most investors who sell in May will be sunk by the taxes due each year on short-term capital gains. Additionally, this strategy is based on a 6-month holding period. If you extend your holding period to 12-months, lo and behold, it turns out that May has historically been a BETTER month than November to initiate your investment strategy! In our analysis, we also review some other trading rules, such as the Super Bowl indicator, which perversely has a better track record than the Sell in May rule, and the U.S. Presidential cycle. We do believe there are disciplined quantitative investment rules that can be applied in the market, and we review two of them in this report. First, our research shows that a cross section of high-quality and dividend-paying stocks have historically outperformed. Additionally, our research confirms that the outperformance of the small and value factors first highlighted by Eugene F. Fama and Kenneth R. French has continued to persist. EXHIBIT 1: GREAT THEORY, BAD STRATEGY Buy and Hold (untaxed) Sell in May (untaxed) Sell in May (w/ 35% tax rate) Annualized returns over total time period Buy and Hold (untaxed) 10.7% Sell in May (untaxed) 10.4% Sell in May (w/ 35% tax rate) 6.8% GROWTH OF $100 $100,000 $80,000 $60,000 $40,000 $20,000 $0 Source: Northern Trust Global Investments, Ibbotson. Data from 11-1-1945 through 12-31-2011. Note: Blue and red dots represent portfolio value after 15% long-term capital gains tax is applied to each strategy. To test the effectiveness of the Sell in May strategy, we modeled the growth of a hypothetical $100 investment from November 1945 to the end of 2011 in three back-tested portfolios: a traditional Buy and Hold portfolio, a Sell in May portfolio, and a taxable Sell in May portfolio. We defined the Sell in May strategy as being 100% invested in the U.S. equity market (S&P 500 Index) from November 1 to May 31 and 100% invested in cash during the rest of the year.

Excess Return (%) The taxable Sell in May portfolio paid a 35% tax rate on all short-term capital gains. As displayed in Exhibit 1, the Buy and Hold portfolio was the best performer, returning 10.7 % annually, while the sell in May portfolio returned 10.4%. The outperformance of the Buy and Hold portfolio is due to the average 2.9% market return from May to November, compared with a 2.2% average return on cash. On a risk-adjusted basis the annualized return over the annual standard deviation the Sell in May portfolio actually outperformed the Buy and Hold strategy; however, a non-taxable investor would have had to explicitly follow the strategy for an extended period to replicate the result. The Sell in May strategy is not nearly as encouraging when taxes are considered. The taxable Sell in May portfolio only returned 6.8% annually, underperforming the Buy and Hold portfolio substantially even after long-term capital gains taxes are deducted on the Buy and Hold portfolio. Based on our calculations, the Buy and Hold portfolio would have needed to lose 1.2% annually between May and November, rather than the positive 2.9% gain observed, for the Sell in May strategy to work. Exhibit 2 reiterates the primary reason behind the success of the Buy and Hold portfolio and the general potential benefits of being invested in the market over the long-term. The data represents the U.S. equity market return less the cash return (going forward, referred to as the excess return) for all 6-month periods starting May 1 going back to 1926. On average, an investor following the Sell in May strategy will forgo a 2.2% excess return. Consequently, a Sell in May strategy has underperformed the Buy and Hold strategy in the long-term. Further, the excess return has tended to increase as interest rates fell. This is apparent when we group the data into quartiles based on the underlying cash returns. When the cash return is between 0.0% and 0.4%, the historical excess return has averaged 8.0%, but it falls to -0.2% when the cash return is above 2.6%. In other words, when interest rates have been low the cost of selling in May and going to cash has been the highest. EXHIBIT 2: IT HAS HURT TO BE OUT OF THE MARKET AVERAGE EXCESS RETURNS AT VARIOUS CASH RETURN LEVELS 8.0 10 8 6 2.2 Across all cash return environments 0.0-0.4% (1st Quartile) Source: Northern Trust Global Investments, Ibbotson. Data from 1926-2011. 0.9 0.4-1.6% (2nd Quartile) 0.2 1.6-2.6% (3rd Quartile) When cash returns are between... -0.2 2.6-7.7% (4th Quartile) 4 2 0-2 While the relationship between the cash return and the excess return reaches the threshold to be considered statistically significant, market returns are dependent on numerous factors and today s low cash rates do not guarantee strong equity performance over the next six months. flexshares.com Page 2 of 11

Average excess return (%) Probability of outperforming cash rate Speaking in statistical terms, our t-statistics on the beta coefficients give high confidence that the cash returns influence excess return variations; but our low r-squared suggests that multiple other factors influence excess returns as well. Intuitively, this is to be expected many variables impact stock market returns over a six month period above and beyond monetary policy; but cash returns do represent a hurdle to surmount when looking for excess returns. In general, while this data may suggest a case could be made for selling in May when cash yields have been high, on average, the investor may be better off in the market (and especially when rates are low). While real-world nuisances, such as taxes and the cost of being out of the market make implementation of the Sell in May strategy difficult, we have yet to answer the question as to whether the Sell in May strategy has any validity as a timing device. Specifically, we need to understand whether there is a strong buy signal attached to November with key measurements being the average returns, the volatility of those returns and the probability of realizing excess returns (referred to as either success rate or hit rate ). We show the key findings in Exhibit 3 and provide a more detailed analysis in the Appendix. The 2.2% figure below should look familiar, as it is the average excess return of 6-month periods starting May 1 (as shown in Exhibit 2). We now compare that 2.2% against the average excess return of 6-month periods starting November 1. By this metric, the November returns look clearly superior as the 5.5% average 6-month excess return is over double that of May. The success rate of November also bests May, albeit by a less impressive margin (71% to 64%). In short, the 6-month returns starting in May and November both beat the cash return on average, but the 6-month returns starting in November do so at a higher level and at a greater success rate. This is where much of the research on Sell in May ends, as this data seems supportive for the market timing efficacy of buying in November vs. May. However, the validity of another timing strategy may help put this in perspective the Superbowl Indicator. According to this theory, you should buy when the National Football Conference (NFC) wins the Superbowl and sell when the American Football Conference (AFC) wins. Using this strategy, you would have seen a 6.2% excess return in years the NFC wins with a remarkable 83% success rate. EXHIBIT 3: THE KEY TO GOOD MARKET TIMING THE SUPERBOWL? 10.0 AVERAGE EXCESS RETURNS & SUCCESS RATES 100% 7.5 64% 71% 83% 75% 5.0 41% 50% 2.5 25% 2.2 5.5 6.2 0.0-1.8 Avg. excess return (left scale) % of time outperforming the cash return (right scale) -2.5 May November NFC AFC 6-month excess returns starting in 6-Month excess returns after Superbowl won by Source: Northern Trust Global Investments, Ibbotson. Data from 1926-2011; Superbowl data starting in 1967. 0% -25% flexshares.com Page 3 of 11

Note: Apologies to non-american football fans. With further time, we would probably be able to uncover a similar market anomaly with world football or cricket! At the risk of stating the obvious, we are not advocating such a strategy. We are pointing out that what appear to be significant statistical relationships can sometimes be a product of pure coincidence. It is helpful to look at the dispersion around the averages to determine how much confidence should be placed in any particular strategy. For example, a noticeable difference in the November average return (vs. the average of all 6-month returns) accompanied by a lower dispersion (vs. dispersion across all returns) would suggest that the month of November is meaningful for some reason. Our tests determined there was a lack of statistical significance (t-stat of 1.6). Other academic research efforts (as noted earlier) have shown some statistical evidence (globally) that the phenomenon exists but have failed to produce a credible reason as to the cause. When directly comparing November and May 6-month returns, we find that the probability of November outperforming in any one year is approximately a 60/40 chance (see first chart in Appendix). What this implies is that the dispersion around the average May returns and the average November returns show quite a bit of overlap. To put this in more concrete terms, we can look at the 2009 experience. In that year, the May-November return of the remarkable rally that started that March was 20%. That 20% return outperformed four out of five November-May returns throughout history. The take-away is that despite May-November having a lower average return, there are several instances where the May returns have competed quite well with the much revered November returns. The analysis we have done so far has only touched on 6-month returns. But most investors have time horizons of a year or longer. Additionally, we are in an election year does that have an impact on the Sell in May strategy? According to our research, it does and the chart below highlights our findings (again, further detail on these returns, along with returns starting in other months can be found in the Appendix). The first set of data below is the same average returns that were presented in Exhibit 3 only now annualized. While the 6-month (annualized) difference is stark, a 12-month time horizon completely removes the discrepancy. Furthermore, when we filtered our data to only incorporate election years, we found that May outperformed November on a 6-month (annualized) basis and also performed well on a 12- month basis. flexshares.com Page 4 of 11

Total Return (%) EXHIBIT 4: GETTING MIXED SIGNALS 11.0 MAY & NOVEMBER EXCESS RETURNS May November 15 12 8.8 8.8 8.6 7.9 6.8 9 4.4 4.8 6 3 6-Month (Ann.) 12-Month 6-Month (Ann.) 12-Month All Data Election Year Source: Northern Trust Global Investments, Ibbotson. Data from 1926-2011; election years represented by 21 episodes. 0 In fact, in a separate study by Ned Davis Research, it was determined that, when looking at only election years, the markets actually bottomed in the month of May. These inconsistencies, while not statistically significant themselves, force us to further question the statistical significance of the original 6-month return patterns. Presumably, if an indicator is truly statistically significant, one should not be able to slice and dice that statistical significance away by using simple filters, such as the presidential election cycle. Incidentally, the presidential cycle represents another market phenomenon. The Presidential Election Theory, as popularized by Yale Hirsch, the creator of the Stock Trader s Almanac, is based on historical observations of stock market returns following a four-year pattern that corresponds to the four-year U.S. election cycle. The theory suggests that, on average, the markets will have the weakest performance in the year following an election than any year of the cycle. We reran the data and found that, from 1928 to 2011, the S&P 500 Index returned an average of 8.6% in the two years following the election, 18.7% in the third year and 11.0% in the election year. The pattern strongly favors stock prices rising the two years prior to a presidential election. However, much like the Sell in May strategy, the statistical significance of this pattern is not overly present. flexshares.com Page 5 of 11

Average return (%) Probability of outperforming cash return EXHIBIT 5: PRESIDENTIAL PUSH 25 20 PRESIDENTIAL CYCLE AVERAGE RETURNS & SUCCESS RATES Avg. return (left scale) 86% % of time outperforming the cash return (right scale) 71% 100% 75% 15 10 52% 50% 50% 5 25% 8.2 9.0 18.7 11.0 0 1st Year 2nd Year 3rd Year 4th Year Year of the presidential cycle Source: Northern Trust Global Investments, Ibbotson. Data from 1926-2011. 0% The theory behind explaining this behavior is tied to the incumbent administration s desire to increase the odds of reelection of not only themselves but their party. Take the bad news (fiscal retrenchment, foreign excursions, etc.) in the first half of the administration and then start priming the pump in the second half. The performance of the economy tends to show similar patterns; although, in contrast to the stock market s behavior, the fourth year of the election cycle has tended to be the strongest economically. While we believe the Sell in May strategy doesn t work, we do think that there are ways to outperform the general market indexes through both active fundamental management and quantitative approaches. Two approaches that our quantitative research team supports are a quality dividend strategy and a small cap/value equity tilted portfolio. We see both strategies having the potential to generate sufficient performance, after expenses and taxes, to support their usage. Our research shows that by combining companies of both high quality and dividend yield, with a portfolio management process that aims to reduce undesirable biases, you may be able to produce persistent excess returns throughout different market cycles. Our focus on quality includes three categories of business performance, including management, profitability and cash. We think these fundamental attributes help measure a company s ability to sustain and grow its earnings and cash flow, which should help deliver shareholder value. Our research shows that the highest quality companies have generally outperformed the market, with lower overall volatility. When analyzing dividend paying stocks, however, we determined that while the top decile in dividend yield did indeed generate excess return, it came with very high volatility. When we analyzed an intersection of high quality and high dividend yield stocks, however, we found that stocks that were in the top 40% of both quality and dividend yield generated excellent performance. By focusing on both high quality and high dividends, we think a portfolio can be positioned to outperform the broad market over the long-term. flexshares.com Page 6 of 11

1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 Growth of $100 EXHIBIT 6: QUALITY AND DIVIDENDS A POWERFUL COMBINATION $50,000 $40,000 $30,000 $20,000 $10,000 $0 GROWTH OF $100 Annualized returns over total time period Intersection Top Quality / Top Yield 19.9% Top Quality 17.7% Top Yield 15.4% Russell 3000 11.3% $39,438 $21,844 $11,279 $3,399 Intersection of Top Quality and Top Yield Top Quality Top Yield R3000 Source: Northern Trust Global Investments, Wilshire Atlas. Data from 1-1-1979 through 12-31-2011. Top Quality is derived from a proprietary Quality Score (QS) that measures the quality of dividend-paying stocks on three categories of business performance: management, profitability and cash stocks with a top quintile rating are determined to be top quality. Top Yield is determined from expected dividend yield at the beginning of the year and separated into deciles with highest dividend-yielding stocks in the first decile. Stocks were selected using a universe of U.S. securities as defined by the Russell 3000 Index. This is a hypothetical illustration, but is based on actual performance. Performance shown in the mountain chart includes the reinvestment of distributions. We have conducted similar quantitative research focused on size and valuation. The work of Fama and French in the early 1990s demonstrated that the smaller a company s market capitalization, and the higher the ratio of the company s book value to market value, the greater its expected return. They found that this return premium provided by size and value were persistent over long periods of time. Our quantitative research analysts have completed their own analysis of the size and value factors, and also believe that they are persistent over time. There is no agreement on why these premia exist. Believers in the efficient markets hypothesis will say they compensate for the extra risk these stocks represent, while others will say that the premium is due to a mispricing of the securities. As with a quality dividend strategy, these strategies will not outperform every year and must be utilized with an intermediate to longer-term time horizon. For example, the top decile of book value to market value (the cheapest) has historically outperformed the most expensive decile 63% of the time on a 1-year basis. If you extend the analysis period to a 5-year period, it jumps to 85% and hits 96% on a 10-year horizon (please see Exhibit 7 on the next page). Patience is certainly key, and those with longer time horizons stand to potentially benefit most. flexshares.com Page 7 of 11

Average excess return (%) Probability of outperformance EXHIBIT 7: SIZE AND VALUE FACTORS SHOW HISTORICAL EFFICACY 10.0 SIZE & VALUE AVERAGE EXCESS RETURNS & SUCCESS RATES Avg. excess return (left scale) % of time outperforming (right scale) 85% 96% 100% 7.5 5.0 69% 62% 63% 57% 75% 50% 2.5 25% 0.0 3.4 2.8 2.6 5.3 4.5 4.6 1-Year 5-Year (Ann.) 10-Year (Ann.) 1-Year 5-Year (Ann.) 10-Year (Ann.) 0% Size: Small Minus Big Value: High Minus Low Source: Northern Trust Global Investments, Ibbotson. Data from 1926-2011. Small is the smallest 50% of stocks by market capitalization and Big is the largest 50% of stocks by market capitalization. High is the top 30% of stocks when ranked by Book Value to Price and Low is the bottom 30% of stocks when ranked by Book Value to Price. The universe includes all stocks listed on the NYSE, AMEX and NASDAQ. CONCLUSION So, should you Sell in May and Go Away? While 6-month returns in May are indeed lower than 6-month returns starting in November, such a strategy would be doomed by taxes, the potential pain incurred by being out of the market (on average) and the wide dispersion of 6- month returns around the averages. Furthermore, our analysis shows that: The underperformance of May disappears in election years Low cash return environments have historically led to notable excess returns on average Seemingly irrelevant data, such as the Superbowl winner, appears to show predictive power Looking at the current situation (presidential election year, low interest rates, New York Giants (NFC team) as current Superbowl champions), one might conclude that this May is not a headwind but a gigantic tailwind! That is not the conclusion of this report. We have demonstrated the statistical vulnerability of the Sell in May strategy, and have shown the shortcomings of its actual implementation. We do see investment opportunities in areas such as quality dividend stocks and small-cap value equities, but not in strategies like Sell in May and Go Away. Past performance is no guarantee of future results. Investing involves risk including the possible loss of principal. Beta (beta coefficient) - Beta represents the systematic risk of a portfolio and measures its sensitivity to a benchmark. Book Value to Price - A ratio used to compare a stock's book value to its market value. It is calculated by dividing the latest quarter's book value per share to the current price of the stock. flexshares.com Page 8 of 11

Excess return (%) Probability density Cumulative probability R-squared- Measures the percentage of a fund's movement that is explained by movements in the market index. Standard Deviation- A statistical measurement of dispersion about an average, which depicts how widely returns varied over a certain period of time. T-statistic (test statistic) - A statistical measure of confidence, it is calculated based on a sample and is the basis for accepting or rejecting a given hypothesis. Russell 3000 Index - Measures the performance of the largest 3000 U.S. companies representing approximately 98% of the investable U.S. equity market. APPENDIX: ADDITIONAL STATISTICAL ANALYSIS 60 STATISTICAL ANALYSIS OF THE "SELL IN MAY" THEORY Distribution of 6-month excess returns by month Standard distribution of Nov. less May excess returns 2.5% 100% May > Nov. 30 0 1.9% 1.3% Normal Distribution Cumulative Probability Probability that May outperforms November: 41% 75% 50% -30 0.6% 25% -60 Individual data points Average 1 2 3 4 5 6 7 8 9 10 11 12 Month of the year (1 = January) Avg. = 3.3% 0.0% 0% -60-40 -20 0 20 40 60 << May Outperf. << Relative Returns >> Nov. Outperf. >> flexshares.com Page 9 of 11

6-Month Period Beg. the 1st of Average Std. Dev. Risk-Adj. Median Hit Rate* Average Std. Dev. Risk-Adj. Median Hit Rate* January 3.8 14.1 0.27 4.4 62% 0.9 13.6 0.07 4.2 67% February 3.8 11.7 0.33 4.0 67% 2.1 8.7 0.24 3.5 71% March 5.0 15.6 0.32 5.0 69% 5.0 7.3 0.68 5.0 81% April 3.7 16.8 0.22 3.4 62% 3.6 8.3 0.44 3.4 71% May 2.2 13.5 0.16 3.2 64% 4.4 11.8 0.38 4.9 76% June 3.3 15.0 0.22 4.8 66% 7.3 16.7 0.44 6.0 86% July 4.3 15.7 0.27 4.7 64% 7.2 16.9 0.43 6.4 76% August 3.9 14.4 0.27 4.3 67% 5.6 13.4 0.42 5.9 76% September 3.1 14.7 0.21 4.3 62% 0.8 15.9 0.05 5.5 57% October 4.7 14.5 0.33 5.1 65% 1.0 13.4 0.08 4.2 62% November 5.5 12.9 0.43 5.2 71% 3.4 11.3 0.30 2.2 57% December 4.8 14.0 0.34 4.4 69% 4.6 14.3 0.32 2.1 57% 12-Month Period Beg. the 1st of STATISTICAL ANALYSIS OF 6- & 12-MONTH EXCESS RETURNS Average Std. Dev. Risk-Adj. Median Hit Rate* Average Std. Dev. Risk-Adj. Median Hit Rate* January 8.1 20.7 0.39 9.8 65% 7.4 17.3 0.43 9.8 71% February 8.3 21.3 0.39 8.8 71% 8.1 17.0 0.47 9.3 81% March 8.6 23.2 0.37 7.1 69% 6.2 20.0 0.31 7.9 71% April 9.0 24.5 0.37 8.5 65% 5.1 18.0 0.29 8.9 71% May 8.8 21.8 0.40 8.9 69% 8.6 20.0 0.43 8.2 71% June 9.2 25.1 0.37 8.2 71% 13.3 30.9 0.43 8.4 76% July 9.3 26.9 0.34 8.3 66% 13.3 37.8 0.35 5.7 71% August 8.4 21.7 0.39 9.3 69% 11.7 23.1 0.51 5.7 71% September 8.0 20.3 0.40 9.4 68% 8.2 20.4 0.40 4.8 62% October 8.0 20.5 0.39 7.5 67% 6.3 19.3 0.33 6.3 57% November 7.9 19.1 0.41 9.6 73% 4.8 18.5 0.26 6.2 62% December 8.0 20.2 0.40 9.6 68% 5.4 23.2 0.23 5.8 57% Source: Northern Trust Global Investments, Ibbotson. Data from 1926-2011 *probability of outperforming cash return Special thanks go to Natalie Sproull for data research. Important Information: Before investing, carefully consider the FlexShares investment objectives, risks, charges and expenses. This and other information is in the prospectus, a copy of which may be obtained by visiting www.flexshares.com. Read the prospectus carefully before you invest. An investment in FlexShares is subject to risk. For a complete description of risks associated with each exchange traded fund, please refer to the prospectus. Foreside Fund Services, LLC, distributor. All Data - Sample Size = 85 Election Year - Sample Size = 21 All Data - Sample Size = 85 Election Year - Sample Size = 21 IRS CIRCULAR 230 NOTICE: To the extent that this message or any attachment concerns tax matters, it is not intended to be used and cannot be used by a taxpayer for the purpose of avoiding penalties that may be imposed by law. For more information about this notice, see http://www.northerntrust.com/circular230. flexshares.com Page 10 of 11

IMPORTANT INFORMATION: This material is for information purposes only. The views expressed are those of the author(s) as of the date noted and not necessarily of the Corporation and are subject to change based on market or other conditions without notice. The information should not be construed as investment advice or a recommendation to buy or sell any security or investment product. It does not take into account an investor s particular objectives, risk tolerance, tax status, investment horizon, or other potential limitations. All material has been obtained from sources believed to be reliable, but the accuracy cannot be guaranteed. PAST PERFORMANCE IS NOT INDICATIVE OF FUTURE RESULTS. Periods greater than one year are annualized except where indicated. Returns of the indexes also do not typically reflect the deduction of investment management fees, trading costs or other expenses. It is not possible to invest directly in an index. Indexes are the property of their respective owners, all rights reserved. No bank guarantee May lose value NOT FDIC INSURED flexshares.com Page 11 of 11