IMPLEMENTING THE DOGS OF THE WORLD STRATEGY USING ETFs

R IMPLEMENTING THE DOGS OF THE WORLD STRATEGY USING ETFs Cantor Fitzgerald Center for Institutional ETF and Indexing Research David Smith, Ph.D Albert Neubert Denise Shelley, Ph.D Research Brief 2.0 November 2014 etf@cantor.com 212 359 8700 In association with

The unpredictability of global equity markets is clearly a challenge for investors that are seeking stable, positive, risk-adjusted returns. Yet certain trends have shown themselves to be rather reliable. As noted in the Cantor Research Brief 1.0, short-term momentum for individual stocks and for markets is an important factor that has even been indexed and in turn, productized in the form of new ETFs. Longer term, however, momentum that may currently be producing superior risk-adjusted returns tends to shift in the opposite direction. This feature of markets is known as return reversal, and is often attributed to investors overreaction to information. (The well-known phenomenon of reversion to the mean is one example of return reversal.) How can investors take advantage of likely return reversals across international markets? The so-called Dogs of the World strategy is an attractive option. In simplified form, the strategy involves buying single-country ETFs associated with last year s worst-performing international markets (the Dogs ) and holding them for 1-5 years the time period during which such markets generally bounce back. Research by Smith and Pantilei (2014), soon to appear in the Journal of Investing, demonstrates the benefits of applying the Dogs strategy over the long run. BACKGROUND For almost 30 years, research has documented return reversal as a powerful force for individual stocks, industries, and international markets. The most influential study of international markets to date is provided by Balvers, Wu, and Gilliland (2000). They follow 18 international markets over a 27-year period and document that return reversals are a regular feature of the investment landscape. They also conclude that reversal strategies require patience, as they often play out over many months. Most readers will recognize a return-reversal strategy as one form of contrarian approach. Contrarian approaches involve going against the market s consensus sentiment. Notwithstanding its rather sour name, a contrarian way of thinking can pay off for several reasons. The world s worst-performing markets in a given year have generally internalized and priced-in negative economic news. In fact, markets may experience overselling in response to bad news and then continue to move downward reflexively in response to later news that in sunnier times would be considered merely ambiguous. It is at just such a time when the economy is stumbling along and policymakers appear to be unable to get anything right when the market often takes its initial steps toward a powerful reversal. And it is much easier to rebound with high-percent returns off the low base left by the previous year s declines. 1 Return reversals for individual stocks have been documented by Fama and French (1988), De Bondt and Thaler (1985), Kim, Nelson, and Startz (1991), and Bali, Demirtas, and Levy (2008). 2 See Gropp (2004) for evidence showing return reversals for industry sectors. 1

DOGS OF THE WORLD Many investors are familiar with the Dogs of the Dow strategy. Michael O Higgins developed that approach as well as the Dogs of the World idea (O Higgins, 2014), which is part of a four-pronged approach presented as the Michael O Higgins Absolute Return Strategy (MOAR). The four portfolio components as seen in Exhibit 1 include investing 25% each in long-term and intermediate-term U.S. Treasuries, 25% in platinum, and 25% in Dogs of the World. All elements of the MOAR can be implemented using low-cost ETFs. However, for the purpose of this Brief, we will focus ONLY on the Dogs of the World component of O Higgins MOAR strategy. Exhibit 1. Components of the Michael O Higgins Absolute Return (MOAR) Strategy Component Weight Representative ETF Ticker Symbol Long-term US Treasuries 25% TLT Intermediate-term US Treasuries 25% IEF Platinum 25% PPLT Dogs of the World 25% Five single-country ETFs Source: O Higgins (2014) In O Higgins version of Dogs of the World, single-country ETFs for last year s five worst-performing markets are bought at the beginning of the current year. They are then liquidated after one year and the next year s five Dogs replace them as portfolio holdings. How does the Dogs of the World portion of the strategy perform? We will assume all returns are in US dollars. Exhibit 2 provides anecdotal evidence, based purely on market index performance, of the return-reversal phenomenon in international markets. The emphasis here is on only the worst-performing market each year, to underscore the effect. Exhibit 2. Subsequent Annual Performance (non-cumulative) of the Worst-Performing Market Worst-Performing Market Subsequent Return Worst-Performing Market Subsequent Return Year Return Market Yr. +1 Yr. +2 Year Return Market Yr. +1 Yr. +2 1988-59% Turkey 547% -5% 2001-41% Egypt 2% 92% 1989-9% Finland -31% -17% 2002-36% Turkey 126% 42% 1990-62% Brazil 172% 7% 2003 20% Finland 7% 18% 1991-45% Indonesia 0% 106% 2004-1% Thailand 9% 12% 1992-45% Turkey 220% -50% 2005-2% Ireland 48% -20% 1993 10% USA 2% 38% 2006-7% Turkey 75% -62% 1994-55% Poland -3% 59% 2007-20% Ireland -72% 13% 1995-31% India -2% 11% 2008-74% Russia 105% 19% 1996-38% Korea -67% 141% 2009-5% Morocco 15% -15% 1997-74% Indonesia -32% 93% 2010-45% Greece -63% 6% 1998-83% Russia 247% -30% 2011-63% Greece 6% 53% 1999-14% Colombia -39% 46% 2012-11% Morocco -3% 2000-62% Indonesia -8% 43% 2013-30% Peru 12%* Notes: All markets returns are for the respective MSCI index s gross return in USD. Shading indicates that the market index outperformed the MSCI All-Country World Index (ACWI) during the year. Also, Peru s return of 12%* (vs. ACWI 5%) is from January through July 2014. Source: MSCI 2

As Exhibit 2 shows, in the first year after experiencing world-benchmark-trailing performance, about half the time, the local market s MSCI index beat the MSCI All-Country World Index (ACWI). But what is particularly notable is the magnitude of outperformance: the Yr. +1 column shows that in six cases, returns have been in the triple digits. The key problem with Exhibit 2 is that most investors would have found it difficult to pursue a Dogs strategy, because of the absence of single-country ETFs for those markets prior to 2001. Since then, the number of markets covered by ETF products has risen to over 40. STRATEGY IMPLEMENTATION USING ETFS Exhibit 3 lists 32 single-country ETFs that have been in existence for a year or more and that have one-way estimated single-day trading costs below 50 basis points for a $20 million trade. In each case, the fund listed has the lowest expense ratio among single-country ETFs for that country. Exhibit 3. Single-Country ETFs Currently Used to Implement the Dogs of the World Strategy Market ETF Ticker Expense Ratio Market ETF Ticker Expense Ratio Australia EWA 0.51% Malaysia EWM 0.51% Austria EWO 0.51% Mexico EWW 0.50% Belgium EWK 0.50% Netherlands EWN 0.50% Brazil EWZ 0.61% Norway ENOR 0.53% Canada EWC 0.51% Russia ERUS 0.61% China MCHI 0.61% Singapore EWS 0.51% Denmark EDEN 0.53% South Africa EZA 0.61% Finland EFNL 0.53% South Korea EWY 0.61% France EWQ 0.51% Spain EWP 0.51% Germany EWG 0.51% Sweden EWD 0.51% Hong Kong EWH 0.51% Switzerland EWL 0.51% India INDA 0.67% Taiwan EWT 0.61% Indonesia EIDO 0.61% Thailand THD 0.61% Israel EIS 0.61% Turkey TUR 0.61% Italy EWI 0.50% UK EWU 0.51% Japan EWJ 0.50% US SPY 0.09% Notes: The ETFs listed above are used in the Dogs of the World strategy. Certain markets in the benchmark MSCI ACWI are excluded from this analysis because a related single-country index ETF does not trade on US exchanges, or because the ETF s estimated one-way trading costs for $20 million exceed 50 basis points. Annualized expense ratios are reported as of August 2014. Prior to introduction of the lower-expense-ratio ETFs listed above, other ETFs were used to provide exposure to the markets of China (ticker: FXI), India (INDY), and Russia (RSX). Source: Morningstar As stated earlier, implementation of the Dogs of the World strategy would involve buying ETF shares in the five worstperforming markets from the previous year and holding for the subsequent 12 months. The selection of the Dogs is made based on the closing price of each index on the last trading day of the year, and rebalancing occurs the next trading day. All returns are assumed to be reinvested, and the five new entrants into the portfolio are acquired in equal weights. The position for any ETF remaining in the portfolio from the prior year is appropriately resized. 3

HISTORICAL PERFORMANCE How would the Dogs approach have worked in practice? We evaluate the strategy starting in 2001, the first year there was a full-year return history for over 20 single country ETFs. Implementation involved taking equally weighted positions in the prior year s bottom five markets, as measured by the corresponding MSCI country index. Returns are computed assuming all trading occurs in the first trading day of the subsequent year, and we further assume that we are able to execute at the closing price. In practice, depending on execution size, consideration should be given to extending the trading horizon to longer than a single day to control transaction costs. However, this extension also brings possible under/out performance due to the lag in transitioning the portfolio. Exhibit 4 depicts the average return for the strategy vs. the ACWI index. All returns are calculated net of management and estimated execution costs. As the exhibit indicates, in the 13+ years between January 2001 and July 2014, the Dogs strategy underperformed the ACWI index in only five of the years. The most dramatic relative underperformance turned out to be in 2013, due to losses in the Indonesia (-24%) and Brazil (-21%) ishares that could not be fully overcome by strong bounce-backs in the ETFs of Spain, Japan, and Israel. The most striking outperformance of the Dogs strategy was in calendar year 2009, led by the returns of Russia (RSX: +122%), Brazil (EWZ: 117%), and Turkey (TUR: +112%), all of which rebounded sharply after the 2008 financial crisis. 2003 was also a big year for the Dogs, led by a 114% return for Brazil (EWZ), which profited from a weak dollar and strong commodity demand. * Through July 2014. 4

The cumulative value of an initial $100 investment is depicted in Exhibit 5. The Dogs portfolio substantially outperformed ($305) the MSCI ACWI benchmark index ($206). And, it should be noted that the MSCI ACWI is being used as a pure benchmark for comparison purposes, sans management and execution costs which would be incurred if replicating the index with an ETF. During this period, the respective compound average returns, net of management and execution costs, were 8.5% for the Dogs versus 5.5% for the ACWI. Even with 2009 removed, the Dogs strategy still generated higher average annual returns than the ACWI. *Through July 2014. Although the observed return difference has been significant, free lunches in financial markets tend to be illusory. The Dogs strategy clearly requires an above-average willingness to endure market volatility. Even though severely troubled markets rebound dramatically more often than not, this hasn t always occurred within the Dogs strategy s 1-year horizon, nor should investors expect the ride within that year to be a smooth one. Investors beware: ongoing pursuit of a deeply contrarian investment strategy virtually guarantees perpetual volatility! To illustrate, the range between minimum and maximum returns across the years is 137 percentage points for the Dogs and 74 percentage points for the ACWI. The annualized standard deviation of returns for the Dogs strategy is 33% vs. 20% for the ACWI. But in this case volatility has an unexpected silver lining: apart from the 2008 experience when global cross-market return correlations neared 1.0, much of the Dogs volatility has been on the upside. This is confirmed by annualized semivariance, a measure of downside volatility, whose square root has been 19% for the Dogs portfolio and 17% for the ACWI. The Sharpe ratios for the observation period were 0.30 for the Dogs portfolio and 0.22 for the ACWI, and the respective Sortino ratios were 0.51 and 0.26. Both Sharpe and Sortino ratios assume a risk-free rate of 3%. 5

Exhibit 6.Trading Cost Estimates (in basis points) based on Current ETF Liquidity Market ETF Ticker Est. Trading Cost for Position Size Market ETF Ticker Est. Trading Cost for Position Size $20M $100M $20M $100M Australia EWA 6.5 15.6 Malaysia EWM 32.7 118.7 Austria EWO 39.5 148.4 Mexico EWW 7.0 24.7 Belgium EWK 21.1 74.7 Netherlands EWN 3.9 9.3 Brazil EWZ 2.4 5.8 Norway ENOR 13.0 37.3 Canada EWC 2.5 4.3 Russia ERUS 6.6 16.6 China MCHI 10.0 20.2 Singapore EWS 23.2 70.8 Denmark EDEN 10.3 30.5 South Africa EZA 7.9 20.9 Finland EFNL 12.2 38.4 South Korea EWY 5.1 35.0 France EWQ 2.1 3.7 Spain EWP 3.3 7.3 Germany EWG 2.3 4.3 Sweden EWD 3.8 8.3 Hong Kong EWH 11.1 28.5 Switzerland EWL 4.7 6.7 India INDA 11.3 38.7 Taiwan EWT 10.0 32.3 Indonesia EIDO 41.0 136.0 Thailand THD 39.8 107.9 Israel EIS 45.7 171.9 Turkey TUR 23.6 69.2 Italy EWI 4.8 9.6 UK EWU 2.2 3.6 Japan EWJ 3.2 6.1 US SPY 0.4 0.6 Notes: Listed above are estimated one-way trading costs (in basis points) for single day execution of trades of $20 million and $100 million. Estimates are as of September 2014. TRADING COST CONSIDERATION Another important implementation concern is the liquidity (and hence transaction costs) of the country-index ETFs. Depending on the initial investment amount, the yearly rebalance can become costly to execute in size due to liquidity constraints. Exhibit 6 summarizes the estimated execution costs for the single-country ETFs that are part of the Dogs of the World portfolio with total portfolio investment amounts of $100M and $500M, respectively. In the exhibit, the $20M column represents the cost of trading 1/5 of a $100M portfolio in a single ETF (because its country index was one of the bottom-five markets), while the $100M column represents the cost of trading that ETF in a $500M portfolio. The costs are estimated based on a full-day execution horizon, and are based on current average daily volume, spread, and volatility metrics. Our estimation process is an oversimplification given that these parameters vary widely across the time period considered in this study. The important point is that the cost of a single day-execution can be large for ETFs with less liquid baskets, and also rise with increasing portfolio notional values, and the execution strategy must be adjusted accordingly. A single-day execution may not always be cost-effective for institutional-sized portfolios, and working with an execution partner that can advise on best execution practices in trading ETF portfolios is important to maximizing the success of strategies such as the Dogs. 6

DOGS FOR 2014 A diverse group of emerging markets exhibiting poor performance in 2013 are the Dogs of 2014. These include Turkey (TUR), Indonesia (EIDO), Brazil (EWZ), Thailand (THD), and South Africa (EZA). China weakness and falling commodity prices in 2013 impacted Latin America and Indonesia, while increased political risk plagued Turkey and Thailand. Thus far in 2014, the Dogs portfolio returns are driven by a strong bounce-back in Indonesia, which at its highest point, had gained up to 30% in 2014. CONCLUSION For institutional investors with relatively long time horizons, the Dogs of the World strategy is capable of harnessing market return reversals and delivering superior cumulative returns relative to the MSCI ACWI benchmark. The ETF structure is ideally designed to implement the strategy, as it provides the individual country exposures in a liquid and easy-to-trade vehicle, and allows seamless initiation and annual rebalancing. In the annual rebalancing of the ETF portfolio, transaction costs must be considered and quantified, and as position sizes grow due to return re-investment, these costs can become large. The Dogs strategy is intuitive in its investment approach: return reversal strategies are conceptually simple and based on the fundamental view that asset prices tend to return to fair value over a long time horizon. REFERENCES Bali, Turan, K. Ozgur Demirtas, and Haim Levy. Nonlinear Mean Reversion in Stock Prices. Journal of Banking and Finance, 32 (2008), pp. 767-782. Balvers, Ronald, Yangru Wu, and Erik Gilliland. Mean Reversion across National Stock Markets and Parametric Contrarian Investment Strategies. Journal of Finance, 2 (2000), pp. 745-772. De Bondt, Werner F. M., and Richard H. Thaler. Does the Stock Market Overreact? Journal of Finance, 40 (1985), pp. 793-805. Fama, Eugene F., and Kenneth French. Permanent and Temporary Components of Stock Prices. Journal of Political Economy, 96 (1988), pp. 246-273. Gropp, Jeffrey. Mean Reversion of Industry Stock Returns in the U.S., 1926-1998. Journal of Empirical Finance, 11 (2004), pp. 537-551. Kim, Myung Jig, Charles R. Nelson, and Richard Startz, Mean Reversion in Stock Prices? A Reappraisal of the Empirical Evidence. Review of Economic Studies, 58 (1991), pp. 515-528. O Higgins, Michael B., Why a Dogs of the World Strategy Is a Winner. Barrons, 20 May (2014). Smith, David M., and Vladimir Pantilei. Do Dogs of the World Bark or Bite? Evidence from Single-Country ETFs, Journal of Investing, forthcoming. 7

The Center for Institutional Investment Management (CIIM) is in the School of Business at the University at Albany, SUNY. The Center s primary mission is to develop and transmit knowledge about institutional investors including their roles, governance, strategies, and performance. http://www.albany.edu/ciim Disclaimer This material is intended for institutional investors only, and as such may not be disseminated, distributed, forwarded or otherwise disclosed to any other party that is not an institutional account, as defined under the Financial Industry Regulatory Authority ( FINRA ) Rule 4512(c). While care has been taken in the preparation of the material presented herein, Cantor Fitzgerald & Co. and its affiliates (collectively, Cantor ) and the University at Albany do not make any representations or warranties, express or implied, as to the accuracy, completeness, or appropriateness of the information contained herein. This document may contain forward-looking statements, which give current expectations of future activities and performance. Any or all forward-looking statements in this material may prove to be incorrect and such statements may be affected by inaccurate assumptions or by known or unknown risks and uncertainties. Cantor and the University at Albany do not undertake any obligation to revise such forward-looking statements to reflect the occurrence of unanticipated events or changed circumstances. Any commentary contained herein may or may not have been prepared by trading desk personnel, who receive compensation based upon various factors, including the overall profitability of Cantor. Cantor and its partners, officers, directors, or employees may sell to or purchase from customers securities, commodity futures, derivatives, or other financial instruments on a principal basis or as agent for another person, and may have interests different or adverse to the interests of the recipient of this communication. Cantor s representatives, other than the author of this communication, may provide oral or written market commentary or trading strategies to Cantor s clients that reflect opinions that are contrary to the opinions expressed in this communication. You are cautioned that there is no universally accepted method for analyzing financial instruments. Past performance is not indicative of future results. Further, Cantor and the University at Albany do not guarantee any results and there is no guarantee as to the liquidity of the instruments described in this communication. As a matter of policy, Cantor and the University at Albany do not give tax, accounting, regulatory or legal advice to clients, and the information contained herein should not be considered to constitute such advice. Cantor and the University at Albany strongly urge that the recipient seek independent legal, regulatory, accounting and tax advice regarding the contents of this communication and make an independent determination regarding the suitability of any strategy or instrument identified herein. This document is not for distribution in the United Kingdom to persons who would be defined as private customers under the rules of the FSA. Cantor and the University at Albany disclaim any and all liability for any loss that may arise from use of the information contained herein. 8