FUNDINDUSTRY INSIGHT REPORT

Transcription

1 FUNDINDUSTRY INSIGHT REPORT JANUARY 213 SECTOR AVERAGES: THE YO-YO EFFECT The temptation to look back at past performance of different sectors in order to decide where to invest next, or when reviewing current fund investments, is hard to resist. The current research paper illustrates the hazards of relying too much on sectors that previously offered the best returns. This analysis is based on the average returns of IMA sectors since 1983, i.e. 3 years of data. 3 IMA Sectors are included, not all of which have been in existence throughout the 3-year period. New and specialist sectors are excluded. The data used is Lipper s calculation of IMA sector indices and accounts for any survivorship bias. The first step is to look at what happens to the IMA sector that has the best average return in each of the past 3 years. Figure 1 plots how each of these top sectors ranked in each subsequent year. For example, the sector that had the best average return in 1983 was ranked sixth in Most recently, the sector that ranked first in 211 (UK Index Linked Gilt) only ranked 27th in 212. Looking at this chart prompts the reader to start nodding as one tries to follow how one year s top sector ranks well in one year, only for the next top sector to rank badly the next year. The average of the points plotted below is 15, in other words, on average, the top sector in one year ranks fifteenth in the subsequent year. The most common subsequent rank is fifth (which occurs three times). Figure 1. Top IMA sector s subsequent rank (1 year periods) Rank for sector that was previously top ED MOISSON Head of UK Research Lipper ed.moisson@thomsonreuters.com SUMMARY This report examines the average returns of IMA sectors over the past 3 years and assesses how the best and worst performing sectors have fared subsequently. The analysis looks not just at 1 year periods, but also 3 years and 5 years, as well as for both sector rankings and average percentage returns. Lastly the correlation across all sectors with their subsequent rankings is also addressed Thomson Reuters. All rights reserved. This report is for informational purposes only, and does not constitute investment advice or an offer to sell or the solicitation of an offer to buy any security of any entity in any jurisdiction. No guarantee is made that the information in this report is accurate or complete and no warranties are made with regard to the results to be obtained from its use. In addition, Lipper will not be liable for any loss or damage resulting from information obtained from Lipper or its affiliates.

2 The same approach can be taken with the bottom ranked IMA sector each year. Figure 2 plots how each of these bottom sectors ranked in each subsequent year. For example, the sector that had the worst average return in 211 (Japan) was ranked ninth in 212. A similar yo-yo pattern is revealed in the chart below. This results in an average of 16 for the points plotted, i.e. on average the bottom sector in any one year subsequently ranks sixteenth. The most common subsequent rank is sixth (which occurs four times). Figure 2. Bottom IMA sector s subsequent rank (1 year periods) 3 Rank for sector that was previously bottom It is also worth rounding out this initial analysis by applying the same approach to the median (middle) sector each year. This serves as a reference point, or control. Figure 3 plots how each median sector ranked in subsequent years. For example, the median sector in 211 (North American Smaller Companies) was ranked 19 th in 212. The chart below shows fewer extremes than in figures 1 and 2, although the average rank is 14 th. The fact that subsequent rankings are more likely to cluster closer to the middle is highlighted by the fact that rankings of fourteen and nineteen both occur three times, the most common subsequent rank, higher (i.e. worse) than the equivalent mode figures for the top (figure 1) and bottom (figure 2) sector analysis above. Figure 3. Median IMA sector s subsequent rank (1 year periods) 3 Rank for sector that was previously the median

3 Comparisons over different periods It is possible to carry out similar analysis of the data for different periods. While figures 1 to 3 above looked at 1 year rankings compared to the subsequent 1 year rankings, the equivalent exercise was also performed for 3 year rankings compared to the subsequent 1 year rankings; 5 years compared to subsequent 1 year; 1 year compared to subsequent 3 years; 3 years compared to subsequent 3 years; 1 year compared to subsequent 5 years; 5 years compared to subsequent 5 years. In order to summarise the findings from this mass of data, averages were calculated for each of the series of data. In figure 4 below, the first blue, grey and orange columns on the left ( 1yr v 1yr ) are simply the averages calculated from figures 1 to 3 above. The remaining six groups of columns in figure 4 similarly reflect the averages calculated for different periods (set out in the paragraph above). Just looking at the average subsequent rankings of the top and bottom sectors (blue and orange columns) over these different periods analysed, one can see that on four occasions the bottom sector s average subsequent ranking was better than that of the top sector, while the reverse was the case on three occasions. Interestingly, the top sector s subsequent ranking was better (i.e. a rank closer to 1 st ) for the 1 year periods assessed, i.e. 1 year compared to subsequent 1 year, 3 year and 5 year rankings. By contrast, the bottom sector s subsequent ranking was better when 3 year and 5 year periods were assessed, i.e. 3 years or 5 years compared to subsequent 1 year, 3 year and 5 year rankings. This is most pronounced when looking at the periods on the right-hand side of figure 4, namely 1 year compared to subsequent 5 years, as well as 5 years compared to subsequent 5 years. For the latter, it is clearly more likely that the bottom sector based on 5 year returns will offer better returns than the top sector in subsequent 5 year periods. Figure 4. Average subsequent rank for top, median & bottom sectors, Average sector ranking yr v 1yr 3yr v 1yr 5yr v 1yr 1yr v 3yr 3yr v 3yr 1yr v 5yr 5yr v 5yr Top sector's subsequent rank Median sector's subsequent rank Bottom sector's subsequent rank

4 So far this analysis has focused on IMA sector rankings, but the equivalent figures were also calculated for average sector returns in percentage terms (bid-to-bid with dividends reinvested, in GBP). The resulting findings presented in figure 5 inevitably underline what was presented in the chart above. Figure 5. Average subsequent return for top, median & bottom sectors, Top sector's subsequent return Median sector's return Bottom sector's subsequent return Average sector return (%) yr v 1yr 3yr v 1yr 5yr v 1yr 1yr v 3yr 3yr v 3yr 1yr v 5yr 5yr v 5yr This additional data helps to quantify the impact of picking different sectors (albeit here, just choosing between those that were previously the best and worst over different periods). Again this chart illustrates the longer the time period, the bigger the potential impact on returns. For example, on average over the past 3 years, picking the worst performing sector over 5 years (if it were possible to invest in this) would have returned 73.2% over the subsequent 5 years, whereas picking the top sector over 5 years would have returned 41.4%. Having said this, if one relied on 1 year sector returns, then picking the worst performing sector would have returned, on average, 41.8% over the subsequent 5 years, while picking the top sector would have returned 7.7%. This latter finding emphasises that if an investor does choose the top sector based on 1 year returns then, on average over the past 3 years, holding that investment has been a much more successful approach than picking the worst or median sector. Interestingly, these 5 year returns also amplify the findings from just a 1 year holding period, where the top sector average also out-performs bottom and median sector averages. By contrast, making an investment based on the sector average over the past 3 years and holding it for just 1 year reveals that investors would have been better off picking the median or even the worst sector, although the same top sector average improves relative to the others when holding for 3 years. Having said this, such a 3 year investment is remarkably similar if it had been based on initial 1 year sector averages.

5 Correlation across all sectors The core of this report is an examination of the subsequent performance of the top and bottom IMA sectors. But the underlying data also enables an assessment of the correlation between the ranking of all sectors in one period with their ranking in subsequent periods. Clearly the chances are small that between twenty and thirty sectors will be ranked in the same way in succeeding years. But the degree to which this changes over time and over different periods provides further useful insights when considering the persistence in sector returns. The three charts below present the resulting correlation coefficients looking at 1, 3 and 5 year data. Figure 6 shows the correlation between 1 year sector rankings and the next 1 year rankings, e.g v Figure 7 shows the correlation between 3 year sector rankings and the next 3 year rankings, e.g v Figure 8 shows the correlation between 5 year sector rankings and the next 5 year rankings, e.g v Figure 6. Correlation across all sector rankings: 1 year periods Correlation coefficient v ' v ' v ' v ' v ' v ' v ' v ' v ' 21 v '2 23 v '4 25 v '6 27 v '8 29 v ' 211 v '12 For all three charts, no persistent correlation can be found between sector rankings. For 1 year returns, it is about as likely that the sectors show a reasonable correlation between one year and the next year (using.6 as a guide), as it is that the sectors are inversely correlated between one year and the next (with -.6 as a guide). But the instances of these stronger correlations decline as the periods assessed are lengthened to 3 years and 5 years. Also, where there are cases of a stronger correlation in sector rankings between two years (figure 6), this drops away in the third year. Figure 7. Correlation across all sector rankings: 3 year periods Correlation coefficient v v v v v v v v v v v v v v v v v v v v v v v v v -12

6 Figure 8. Correlation across all sector rankings: 5 year periods Correlation coefficient v v v v v v v v v v v v v v v v v v v v v 8-12 Concluding comments The variability in the ranking of IMA sector returns over the past thirty years is such that the average ranking is close to the mid point (of 3 sectors) whether one chooses the top, bottom or middle sector from the previous year. There are variations in the findings when looking at different time periods, albeit that average rankings (see figure 4) all range between 12 th and 18 th. The difference between top and bottom sectors rankings are most pronounced when looking at these sectors subsequent 5 year returns, although to a certain degree these patterns can already be seen when looking at 1 year returns. The hazards of an over-reliance on past sector returns are applicable for investors in both index trackers and actively managed funds. The former because there is no escape from investors asset allocation decisions and the latter because even good managers are still often beholden to the prevailing winds that may buffet a sector. Finally, thanks are due to my colleagues Ran Vispap-Rich and Noora Vainio for their invaluable help in preparing this research. REUTERS / Pawel Kopczynski

7 APPENDIX Figure 9. Top and bottom IMA sectors Rank based on average of funds' 1 year % performance, bid-to-bid, dividends reinvested, in GBP. Year Top sector Bottom sector 1983 Japan Corporate Bond 1984 UK Equity Income & Growth Technology & Telecoms 1985 Europe Excluding UK Asia Pacific Including Japan 1986 Asia Pacific Including Japan North American Sm Companies 1987 High Yield European Smaller Companies 1988 Asia Pacific Excluding Japan Corporate Bond 1989 Global Emerging Markets Corporate Bond 199 Money Market Japan 1991 North American Sm Companies Property 1992 North American Sm Companies Property 1993 Asia Pacific Excluding Japan Money Market 1994 Japanese Smaller Companies Global Emerging Markets 1995 Technology & Telecoms Global Emerging Markets 1996 Property Japanese Smaller Companies 1997 North America Japanese Smaller Companies 1998 Technology & Telecoms Global Emerging Markets 1999 Japanese Smaller Companies Global Bonds 2 Property Japanese Smaller Companies 21 Corporate Bond Technology & Telecoms 22 UK Index Linked Gilt Technology & Telecoms 23 Global Emerging Markets UK Gilt 24 European Smaller Companies Technology & Telecoms 25 Japanese Smaller Companies Money Market 26 European Smaller Companies Japanese Smaller Companies 27 Asia Pacific Excluding Japan Japanese Smaller Companies 28 Global Bonds UK Smaller Companies 29 Global Emerging Markets Japan 2 North American Sm Companies Money Market 211 UK Index Linked Gilt Global Emerging Markets 212 UK Smaller Companies UK Index Linked Gilt Note: Newer and specialist sectors excluded for the purposes of the current report: Absolute Return, China/Greater China, Global Equity Income, Mixed Investment -35% Shares, Pensions, Protected, Short Term Money Market, Unclassified, UK Zeros.