Empirical evidence of excess returns in emerging markets
Jorrit de Jong
Groningen, December, 2010
Master Thesis Business Administrator - Finance Rijksuniversiteit Groningen
1
Empirical evidence of excess returns in emerging markets
Jorrit de Jong
Abstract
This thesis examines emerging markets empirically by looking at 47 emerging countries over a period of 1976 till 2010. We found higher risk adjusted excess return in emerging markets than in developed markets (G7). Furthermore, an upward trend was observed in the correlation between the emerging markets excess returns and various global influences. Moreover, we found a very high correlation between developed markets and emerging markets in recent years. This may be a sign of further integration of emerging markets to the global markets.
Acknowledgement
2
Table of contents
1. Introduction ...3
2. Methodology ...6
3. Data Description ...7
3.1 Stock market data...7
4. Results ... 12
4.1 Global influences on emerging markets ... 14
4.2 Secular markets characteristics of emerging markets ... 17
3
1. Introduction
Emerging markets received a lot of attention in the last decade. But is investing in these markets promising? Shifting away from relatively safe developed markets to emerging markets boosts the risk of an investment portfolio. Both Salomons and Grootveld (2003) and Shackman (2006) showed an increase in both risk and returns in emerging markets compared to developed markets. In a historical perspective, investors receive higher returns in the riskier emerging markets.
This thesis is meant as a contribution to the discussion about returns in emerging markets. It includes more countries and a longer timeline than previous studies. Furthermore, this thesis investigates the stability of correlations of emerging markets with some possible global influences and secular market characteristics of returns in emerging markets. The thesis gives an empirical overview of excess return in emerging markets over time. Excess return including risk adjusted excess return will be compared between developed and emerging markets. In the quest to unravel parts of the equity premium puzzle this thesis will also look deeper into some possible drivers of excess returns in emerging markets.
Our main research question is: Have excess returns in emerging markets changed in recent years? This question can be divided into the following subquestions: Have emerging markets become less or more cyclically driven? Can global liquidity be linked to emerging markets? Do commodities have an impact on excess return in emerging markets? Can the value premium or size premium of Fama and French (1996) be linked to emerging markets? Can change in excess returns be linked to emerging markets health?
Over the period from 1802-1990, Siegel (1992) found an excess return of equities in de US of 5.3% per annum. The positive excess return can be seen as a reward for additional risk of stocks over risk free assets. There is a vast literature on developed markets, and positive excess returns were observed in a number of developed countries (Goetzmann and Jorion, 1992, Dimson, 2003). In the last decade emerging markets have gained more interest.
4
narrower definition of the Morgan Stanley Capital International (MSCI) Emerging Markets Index. They used a sample of the 10 countries out of 25 with the highest market capitalization. Kohers et al. (2006) also followed the MSCI Emerging Markets Index in defining emerging markets. Furthermore, Goetzmann and Jorion (1999) argued that emerging markets are re-emerging markets, with active equity markets in the 1920s. The country definition of emerging markets may therefore not be stable over time.
Salomons and Grootveld (2003) studied 24 countries in emerging markets and included dividends in their calculations. They found the excess return in emerging markets to be higher than in developed markets. Shackman (2006) also looked at total returns and use only countries with a data history of more than 20 years. He studied 20 countries in emerging markets over a longer time frame. He found that emerging markets do have higher excess returns than developed markets, but, when adjusted for risk, developed markets have higher returns. Kohers et al. (2006) studied 25 emerging markets over a period of more than ten years. They found the standard deviation in emerging markets to be higher than in developed markets. Besides, they found risk-averse investors compensated for the higher risk with higher returns. Shackman (2006) mentioned the limited power of comparing only the excess returns of developed markets and emerging markets. He argued that the equity volatility is also part of the equation. The equity premium puzzle is therefore really a Sharpe ratio puzzle (Sharpe, 1966).
Salomons and Grootveld (2003) did not only calculate the Sharpe ratio but the Sortino ratio as well (Sortino and van der Meer, 1991). They argued that in a standard deviation framework model (like the Sharpe ratio) large positive outcomes are treated as equally risky as negative ones. Positive outcomes, however, should be regarded as a bonus and not a risk. Using the Sortino ratio returns were adjusted for downside risk. Salomons and Grootveld (2003) found more downside risk in emerging markets compared with developed markets.
5
emerging markets can be treated as a distinct asset class by global investors, which is in line with the spirit of this study.
Salomons and Grootveld (2003) found some evidence that excess returns in emerging markets behave cyclically. They found that emerging markets are correlated to the global economic cycle, based on OECD industrial production. Neumeyer, Pablo, Fabrizio (2004) found evidence that business cycles in emerging markets are more volatile than in developed markets. Calvo, Leiderman and Reinhart (1993) studied Latin America and suggested global interest rates (global liquidity) as an important factor to excess return in emerging markets. The IMF (2010) report stated a negative relation between global liquidity and real interest. It also positively associated global liquidity with equity returns.
In Johnson and Soenen’s article (2009) the relation between stock markets and commodity prices in South America was studied. A positive relation was found between commodities and stock returns after controlling for exchange rates, interest rates and the US stock market. Anderson (2009) argued that the growth of emerging markets drove up commodity prices, although Johnson and Soenen (2009) found no lead or lag relation of any significance in their sample. Therefore, in this thesis we will look into the global economic cycle, global liquidity and commodity prices as global influences on emerging markets.
Rouwenhorst (1999) concluded that return factors in emerging markets are qualitatively similar to those in developed countries. He found that value and size premiums also exist in emerging markets. Drew and Veeraraghavan (2003) showed evidence of the existence of the three-factor model of Fama and French (1996) in emerging markets. The health of emerging markets may have a certain effect on emerging markets. Salomons and Grootveld (2003) gave an overview of the Latin crisis, with large negative excess returns caused by a country-level debt crisis in 1982. It is argued by Anderson (2009) that “balance sheets” of emerging markets are a crucial factor in explaining recent high growth. Poor health of the sovereign debt may increase the riskiness of the emerging markets, which, in its turn, may lead to higher compensation for investors in these markets. We examine if excess returns can be explained by other risk factors. Therefore, in this thesis we will look into the three-factor model of Fama and French (1996) and the health of emerging markets as secular markets characteristics.
6
emerging markets and global liquidity (real interest), leading indicators and commodities. Moreover we found higher risk adjusted returns in emerging markets than in developed markets.
Methodology of this thesis is explained in section 2. In section 3 the data description is explained. In section 4 the results are presented. The last section concludes.
2. Methodology
The analysis of the returns in emerging markets consisted three parts: in the first risk adjusted emerging markets returns are compared with those of developed markets. Second we looked at the correlation of possible global time-variation influences on excess returns in emerging markets. Those possible influences are the global economic cycle, real interest and commodity prices. In the third, possible influences on emerging markets are discussed; in contrast to the previous part we looked at characteristics of the emerging market itself.
The three-factor model of Fama and French (1996) was used to examine the value premium or the size premium as explanatory variable for the different characteristics of the emerging market. The health of emerging markets was examined to see whether or not it could explain the higher compensation that investors receive in emerging markets. We will now give an overview of the methods used in the three different parts of the thesis.
1. Risk adjusted excess return: The comparison between emerging and developed markets was made by the Sharpe ratio and the Sortino ratio. Nevertheless, we also looked at the absolute differences between emerging markets and developed markets and tested for differences. Additionally, we tested for changes in the level of excess return using the method of Bekaert (1995), splitting up the data into subsamples. The subsamples were split by high impact financial events: black Monday in 1987, the dotcom crisis in 2001. We tested for differences between developed markets and emerging markets and between sample periods.
2. Global influences on emerging markets: We used a twelve months moving average filter to investigate the correlation of possible global time-variation influences on excess returns in emerging markets. Furthermore, we looked at rolling correlations. The global economic cycle, global liquidity and commodities prices were examined. All correlations were tested though the Spearman rank-order correlation to show stability of the correlation.
7
of the emerging market. It may explain differences in characteristics of the emerging market by the value premium or the size premium. We used the following regression:
Where is the term for developed markets excess return.
is the term for excess return of an US HML stock portfolio. is the term for excess return of an US SMB stock portfolio.
The health of emerging markets was used to see if it could explain the possible higher return investors receive in emerging markets. We used twelve months moving average filter to investigate the correlation between the difference between emerging and developed markets and the relative differences of the emerging markets investment grade index.
3.
Data Description
All elements needed to calculate the excess returns for emerging and developed markets are described. The data sources of all variables used in this thesis are described in Appendix 2. 3.1 Stock market data
This study used an index for the emerging markets and focuses on emerging markets in contrast to developed markets. This universal approach is most appropriate for a US investor (global investor) who is able to invest in many countries in the world. The stock market data used are float adjusted market-capitalization based indices of the stock markets of 54 countries. We collected data on 47 emerging markets and 7 developed markets (G7). The countries are shown in Table 11. We only used the G7 countries for the developed market, since they serve as a benchmark to the emerging markets, which is consistent with Salomons and Grootveld (2003). The definition of emerging market in this study is the same as the MSCI market definition of Augustus 2010. By selecting the 47 countries we tried to capture the countries that investors in emerging markets tend to focus on. The index returns are monthly total returns in US dollars. The data had a time span from February 1976 till August 2010. Some of the emerging market countries have relatively small market-capitalization, are illiquid or have legal constraints for foreign investors. Consequently, very strong conclusions cannot be drawn.
1
8
Two equally weighed indices were made from the individual data series per country: one for the emerging market and one for the developed market. This enabled us to speak about emerging markets in a broader context, without going into country specific details.
Table 1. Index constituent. All countries used in this study are shown. In the
first column we show the developed (G7) countries of the developed index. In the other four columns the emerging markets countries of the emerging index are shown.
Developed Index
Emerging
Index
G7 Asia Latin America
Middle East & Africa
Eastern Europe
Canada China Argentina Bahrain
Czech Republic
France India Brazil Egypt Hungary
Germany Indonesia Chile Jordan Poland
Italy Korea Columbia Kenya Russia
Japan Malaysia Mexico Kuwait Turkey
United Kingdom Pakistan Peru Mauritius Slovakia
United States Philippines Venezuela Morocco Bulgaria
Sri Lanka Nigeria Croatia
Taiwan Oman Estonia
Thailand Qatar Kazakhstan
Vietnam Saudi Arabia Lithuania
South Africa Romania
Tunisia Serbia
UAE Slovenia
Ukraine
The return characteristics of all countries are shown in Appendix 3. Following Shackman (2006) we calculated Sharpe ratios for each individual country. It can be noticed that Argentina is an outlier. This country has a high average monthly return of 3.39% and in the first period a high volatility with a maximum monthly return of 178% in 1989. Not all emerging markets perform well over time, for example, China have a low monthly total return of only 0.55%. Figure 1 shows the distribution of the emerging and developed market index returns over the period of 1976-2010 years.
9
Fig. 1. Distribution of the emerging and developed markets. Monthly equal index total returns
based on the period of 1976 till 2010. Both graph have similar scale and are in US dollars.
Panel A: Emerging markets Panel B: Developed markets
Table 2. Statistics for developed and emerging markets. All
figures are presented of both emerging and developed market monthly total return index, nominal in US dollars. The standard deviation (std dev) is calculated monthly over the entire sample period. When calculating the Sharpe and Sortino ratios, 10 years treasury total returns of the US are used. The Sharpe and Sortino ratios are annualized. The last three variables give information about the normality of the distribution, where Jarque-Bera is the test for non-normality
Equal weighted
Return Developed index Emerging index
2/1976 - 8/2010 2/1976 - 8/2010 Mean 1.02% 1.84% Median 1.19% 1.92% Maximum 15.31% 21.02% Minimum -20.73% -26.92% Std. Dev. 4.81% 5.52% Sharpe ratio 0.20 0.69 Sortino ratio 0.35 1.26 Skewness -0.57 -0.55 Excess Kurtosis 1.93 2.91 Jarque-Bera 84.01* 161.95* * Significant at 1%
10
result of 0.35, but they found a Sortino in emerging markets of 0.87, where we found a higher Sortino ratio of 1.26. This difference may result from the use of a different risk free rate, a longer time frame and more countries in our sample.
Shackman (2006) used real returns and real discount rates to calculate the Sharpe ratio. We followed this approach and found that the Sharpe ratios for the developed and emerging markets did not change much. His study used a cross-sectional average on the Sharpe ratio of both emerging and developed markets2. By employing this method we found a Sharpe ratio of 0.16 for developed markets and 0.30 for emerging markets. This method places a little more emphasis on recent observations, as a result of the availability of data as shown in Figure 10 of Appendix 1. Salomons and Grootveld (2003) found Sharpe ratios of 0.24 in developed and 0.57 in emerging markets.
The Ljung and Box (1978) Q-statistics were used to determine the amount of autocorrelation of emerging and developed markets. We found that emerging markets have one month autocorrelation with significant Q-statistic at 1%. Developed markets do not have significant autocorrelation. The first order autocorrelation of the emerging markets can be an indication of liquidity problems in these markets (Khandani and Lo, 2009).
We examined differences between emerging markets and developed markets. Furthermore, we followed Salomons and Grootveld (2003) and tested for changes in the level of excess return where the method of Bekaert (1995) was used. We split the data into subsamples. The subsamples were split by high impact financial events: black Monday in 1987, the dotcom burst in March 2000. For the dotcom burst we chose March because that was the month where the Nasdaq was all time high. We tested for differences between developed markets and emerging markets and between sample periods, presented in Table 3. No significant conclusions can be drawn based on this subsamples test. The differences between developed markets and emerging markets and between sample periods turned out to be not significantly different.
2
11
Table 3. Sub samples of the equity risk premium. We tested for
differences between developed markets and emerging markets and between sample periods. The subsamples were split by high impact financial events: black Monday in 1987, the dotcom burst in March 2000. T-Tests for Equality are the tests performed on the horizontal rows (right) between subgroups and performed on the vertical rows between time periods.
Equal weighted excess return
Emerging index
Developed
index Test for Equality
Mean (1976-1987) 1.48% 0.80% t=0.79
Mean (1987-2000) 0.96% 0.19% t=1.09
Test for Equality t=0.64 t=0.80
Mean (1987-2000) 0.96% 0.19% t=1.09
Mean (2000-2010) 0.77% -0.28% t=1.20
Test for Equality t=0.23 t=0.64
We tested excess returns for equality over the entire period and found that the t-test score is 1.74 and is significantly different at 10%. The Wilcoxon test, to test the median, had a value of 1.89 and was also significantly different at 10%. We also performed the Wilcoxon test because the emerging and developed excess returns were not normally distributed.
Fig. 2. Correlation of returns in emerging and developed markets. We present the 5y rolling
correlation between emerging and developed markets over time. Based on monthly equal index total returns on the period of 1976 till 2010. Where the correlation of the rolling period of 5 years is calculated again after each month move till 2010. The steps we take from the 5 years rolling period is 1 month. Therefore we calculated the period of 1981 till 2010.
12
countries that had covered the entire period, the rolling 5 years correlation remained the same with an increase from 0.148 to a steady 0.908. This higher correlation might be a result of global market integration. If we tested the correlation between emerging markets excess returns and developed markets excess returns we found a correlation of 0.74. As an alternative test we used the Spearman rank-order correlation which resulted in a value of 0.68.
Fig. 3. Sharpe ratio’s in emerging and developed markets. We show the 20y rolling
annualized Sharpe ratio’s in emerging and developed markets. Based on monthly equal index total returns on the period of 1976 till 2010. When calculating the Sharpe ratios, 10 years treasury total returns of the US were used. Where the correlation of the rolling period of 20 years is calculated again after each month move till 2010. The steps we took from the 20 years rolling period is 1 month. Therefore we calculated the period of 1996 till 2010.
Figure 3 shows that the Sharpe ratios in the emerging and developed markets was not constant, but dependent on the time frame calculated. The period 2008 until 2009 had a negative Sharpe ratio in developed markets because of the risk free rate, the 20 years rolling average returns are positive over the entire period. The rolling Sharpe average of developed markets was 0.20 and the average for the emerging market was 0.63.
4. Results
Historically emerging markets tend to have a strong local character (Rouwenhorst 1999). In the last decade much has changed and therefore we took another look at global risk factors and other global factors that may influence emerging equity returns.
13
Figure 4 gives an overview of the period from 1976 till 2010 using a twelve month moving average, in which the Industrial production tends to follow the leading indicator.
Fig. 4. Industrial production and Leading indicators. The figure gives the Twelve month
moving average over time. The industrial production and the leading indicators are from the OECD and calculated of the G7.
We used the US Real interest as a proxy for the global liquidity. Markets participants use this definition of global liquidity. The real interest was calculated by the Fisher equation.
For commodity prices the monthly CRB Spot Index was used as proxy for commodity prices. In the period that we studied the monthly average was 0.26%, the monthly highest return in February 2008 was 9.29% and the lowest was -20.1% in November 2008. For the commodity index we found a standard deviation of 2.64% over the entire period.
We used Monthly US Historical Benchmark of the Value premium (HML) and Size premium (SMB) from 1976 till 2010. Our strong assumption was that trends in value premium and the
size premium in the US are observable worldwide. The size premium (SMB) and the value
premium (HML) were defined equally as in the paper of Fama and French (1996).
The health of emerging markets was defined by the country credit rating.3 This rating varied from AAA for the least risky countries to CCC for countries almost in default. Where investment grade is defined as a rating higher than BBB. In January 1995, only 47% of all emerging countries in our sample had an investment grade rating. This improved to 69% in Augustus 2010 for all emerging countries in our sample. In Figure 5 the investment grade of
3
14
emerging markets over time is given. It should be noticed that our measure for health is very broad; it can also be seen as strength of countries in emerging markets.
Fig. 5. Investment grade of emerging markets. We present the equal index of investment
grade based on credit rating of the emerging markets out our sample over a period of 1995 till 2010. The rating varies from AAA for the least risky countries to CCC for countries almost in default. Where investment grade is defined as a rating higher than BBB.
4.1 Global influences on emerging markets
15
Fig. 6. Emerging markets excess returns and leading indicators. We show the twelve month
moving average of emerging markets excess returns and the leading indicator. The leading indicator is from the OECD and calculated of the G7. The emerging markets returns are total returns of an equal weighted index in US dollars. When calculating the excess returns, 10 years treasury total returns of the US were used. The left vertical axis is the scale of the emerging markets and the right vertical axis is the scale for the leading indicator.
For considering the link with global liquidity, we used the market practitioners’ definition of global liquidity, the real interest. As can be seen in Figure 7, the correlation of the emerging markets are above those of the developed markets in most of the period; only the subperiod 2002 to 2008 is reversed. In November 2008 there was a large break in the data; the real interest was high and the emerging and developed markets were quite negative. The correlation over the entire period for the emerging markets was -0.64 and for the developed markets -0.71. This negative relation is in line with results of the IMF (2010) which stated a negative relation between global liquidity (M2)4 and real interest. The report also positively associated global liquidity with equity returns. When we tested the correlation between emerging markets excess returns and real interest returns we found a correlation of -0.64 and a Spearman rank-order correlation of -0.62.
4
16
Fig. 7. Correlation of excess returns and the real interest. We show the 5y rolling correlation
between excess returns and the real interest. Based on monthly equal index total returns based on the period of 1976 till 2010. When calculating the excess returns, 10 years treasury total returns of the US were used. Where the correlation of the rolling period of 5 years was calculated again after each month move till 2010. The steps we took from the 5 years rolling period is 1 month. Therefore we calculated the period of 1981 till 2010. Real interest is calculated though the Fisher equation based on US inflation and 10 years treasury bonds returns.
17
Fig. 8. Correlation of excess returns and commodity returns. We show the 5y rolling
correlation between excess returns and commodity returns. Based on monthly equal index total returns based on the period of 1976 till 2010. When calculating the excess returns, 10 years treasury total returns of the US were used. Where the correlation of the rolling period of 5 years is calculated again after each month move till 2010. The steps we take from the 5 years rolling period is 1 month. Therefore, we calculated the period of 1981 till 2010.
4.2 Secular markets characteristics of emerging markets
To test if there is a link with the value premium or size premium of Fama and French (1996) we did a regression5. We adopted the regression format as presented in the method section. We found an adjusted R-squared of 57%, with a Durbin-Watson of 1.46. All terms of the regression were significant at 1%. The regression had a Jargue Bera of 3.13, normally distributed. Additional calculations and graphs of the rolling calculations are presented in Appendix 4. We saw the R-squared of the model become higher over time.
To answer the question if the health of emerging markets can be linked to excess returns, we examined the relation and correlation between investment grade index and the emerging markets returns. Figure 9 shows that the relation between emerging markets returns minus developed markets returns to investment grade index of emerging markets is not a constant. The relation amplitude seems to decline over time. Simply looking at the emerging excess returns correlation to relative differences of the emerging markets investment grade index changes gave a low correlation. But when we tested the correlation between emerging minus developed markets returns and the relative differences of the emerging markets investment grade index we found a correlation of 0.16 and a Spearman rank-order correlation of 0.12.
5
18
When we broke down the sample period in two, we found a correlation in the first half (n=93) of 0.16 and latter half(n=93) of 0.20.
Fig. 9. Difference between emerging markets and developed markets returns and the change of the investment grade. We show the twelve month moving average of both variables.
The emerging and developed markets returns are total returns of an equal weighted index in US dollars. The investment grade is the change of an equal index of an investment grade dummy of the emerging markets over a period of 1995 till 2010. Were for the investment grade we take the change compared to the previous month investment grade.
5. Conclusion
The results of this thesis show that excess returns of emerging markets are higher than in developed markets, which is found in similar studies. Furthermore, we found risk adjusted excess return of emerging markets are higher, measured by the Sharpe and Sortino ratio. But these results should be interpreted with caution, because we showed that Sharpe ratios vary over time. Still, by using 20 years rolling Sharpe ratios over more than three decades, all observed ratios of emerging markets are above the developed markets ratios.
The correlation of emerging markets to the developed markets have increased over time. In the most recent period we even found a very high 5 years correlation. This higher correlation might be a result of high market integration. Beside this result, we found first order autocorrelation in emerging markets in contrast to developed markets. This autocorrelation could be a sign of illiquidity or investment barriers in emerging markets.
19
behavior which is observed in previous studies in emerging markets is also observable over a longer time period and in more countries as this study does. Emerging markets became even more cyclically driven than developed markets, measured by the correlation in the latest period. We tested the correlation of global liquidity as measured by the real interest returns and commodity returns both significant with excess returns. The real interest returns had a high correlation and the commodity returns a substantial correlation with excess return in emerging markets over the entire period.
In an attempt to explain the differences between emerging markets and developed markets we looked at the factor model of Fama and French (1996). Directly applying the three-factor model on excess return in emerging markets gave high significant terms and adjusted R-squared. Although the terms were all significant, we made a strong assumption of worldwide observable trends. The excess return in emerging markets is a complicated puzzle, therefore we only want to suggest value premium and size premium as explanatory variables for future study.
We looked into the health of emerging markets measured by the country rating. We tested the correlation between emerging minus developed markets returns and the relative differences of the emerging markets investment grade index. We found a low correlation. It appeared that the health of the emerging markets partly, but not strongly, explains the gap between emerging markets and developed markets.
20
Appendix 1:
1. Stock market data
1.1 Country total return data series
For all countries the MSCI indices and the S&P/IFC indices were extracted from DataStream and combined to obtain the longest possible data series per country6. Unfortunately, this process was necessary because most of the emerging markets indices of countries did not cover the entire period. The following rules were applied:
1. The first data point index is the starting index, where, if equal, MSCI prevails. 2. If an index ends, the next longest surviving index will continue the series.
Data from the developed markets were available for all seven counties for the entire sample period. Using the rules as mentioned above, only MSCI data were used. Figure 10 shows the number of available indices of emerging markets over time, after applying the combining process.
Fig. 10. Number of available indices of emerging markets. In the figure we show the number
of countries that have been used in the equal weighted index of emerging markets over time.
It should be noticed that MSCI and S&P/IFC stopped the Venezuela index series in January 2008 as this became non-investable.
1.2 Risk free rate
For the very short maturity, the monthly FED fund rate was used and for the longer maturity, the monthly 10 years US treasury total return was used. The data were extracted from the website of the American Federal Reserve System (FED) and DataStream.
6
21
Appendix 2:
1. Data sources:
K. R. French website: http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/index.html American central bank (FED) website: http://www.federalreserve.gov/
Robert J. Shiller website: http://www.econ.yale.edu/~shiller/data.htm Bloomberg website: http://www.bloomberg.com/
Standard and Poor’s website: http://www.standardandpoors.com/
Moody’s website http://www.moodys.com
2. Explanatory variables
2.1 The health of the emerging markets
The health of emerging markets was defined by the country credit rating from Standard and Poor’s and Moody’s with a time span of January 1995 until Augustus 2010. This timeframe is shorter because the availability was limited.
2.2 Value premium and the size premium
The US data for the value premium (HML) and Size premium (SMB) was extracted from the data library of Professor Kenneth R. French’s website.
2.3 Real interest
22
2.4 Commodity prices returns
For commodity prices the monthly Commodity Research Bureau Spot Index was used as proxy for commodity prices. The index has a time span from 1976 until 2010 and was extracted from DataStream.
2.5 Production and leading indicators
23
Appendix 3:
Table 4. Distributional characteristics of monthly returns of developed markets, denominated in US$
Canada France Germany Italy Japan
United Kingdom
United
States MSCI World G7 (Equal)
2/1976 - 8/2010 2/1976 - 8/2010 2/1976 - 8/2010 2/1976 - 8/2010 2/1976 - 8/2010 2/1976 - 8/2010 2/1976 - 8/2010 2/1976 - 8/2010 2/1976 - 8/2010 No. 415 415 415 415 415 415 415 415 415 Mean 1.06% 1.12% 1.03% 1.00% 0.87% 1.11% 0.96% 0.92% 1.02% St. Dev. 6.04% 6.66% 6.55% 7.55% 6.37% 5.81% 4.58% 4.44% 4.81% Skewness -0.48 -0.15 -0.26 0.21 0.27 -0.02 -0.57 -0.59 -0.57 Excess Kurtosis -0.57 -1.82 -1.62 -2.17 -2.28 -1.53 -0.71 -1.00 -1.07 Median 1.29% 1.22% 1.22% 0.82% 0.69% 1.05% 1.19% 1.22% 1.19% Min -25.77% -23.18% -22.03% -22.60% -19.38% -21.53% -21.22% -18.74% -20.73% Max 23.30% 26.52% 23.78% 30.99% 24.26% 22.66% 15.79% 14.92% 15.31% Sharp 0.19 0.20 0.16 0.12 0.07 0.22 0.17 0.15 0.20
Table 5. Distributional characteristics of monthly returns of emerging markets, denominated in US$
China India Indonesia Korea Malaysia Pakistan Philippines Sri Lanka Taiwan Thailand Vietnam Asia (Equal) EM (Equal)
24
Table 6. Distributional characteristics of monthly returns of emerging markets, denominated in US$
Argentina Brazil Chile Columbia Mexico Peru Venezuela
Latin America (Equal) EM (Equal) 2/1976 - 8/2010 2/1976 - 1/2008 2/1985 - 8/2010 2/1993 - 8/2010 2/1995 - 8/2010 2/1985 - 8/2010 2/1976 - 8/2010 2/1976 - 8/2010 2/1976 - 8/2010 No. 415 415 415 307 415 211 276 415 415 Mean 3.39% 2.13% 2.14% 2.21% 1.84% 1.98% 1.86% 2.37% 1.84% St. Dev. 22.50% 14.85% 9.49% 8.86% 11.04% 9.85% 13.78% 8.94% 5.52% Skewness 2.49 0.37 0.87 0.59 -0.86 0.16 0.31 0.53 -0.55 Excess Kurtosis 11.02 -1.40 1.46 -0.85 1.22 -0.72 0.06 -0.70 -0.09 Median 1.53% 1.03% 1.51% 1.89% 2.20% 1.74% 1.02% 2.18% 1.92% Min -64.95% -56.89% -28.03% -25.33% -59.32% -33.62% -49.79% -31.59% -26.92% Max 178.11% 57.53% 62.86% 37.34% 39.60% 40.30% 59.58% 42.30% 21.02% Sharp 0.41 0.32 0.51 0.57 0.35 0.50 0.28 0.63 0.69
Table 7. Distributional characteristics of monthly returns of emerging markets, denominated in US$
Bahrain Egypt Jordan Kenya Kuwait Mauritius Morocco Nigeria Oman Qatar Saudi Arabia
25
Table 7. continued
South Africa Tunisia UAE
Middle East & Africa (Equal) EM (Equal) 2/1993 - 8/2010 7/2004 - 8/2010 7/2002 - 8/2010 3/1978 - 8/2010 2/1976 - 8/2010 No. 211 74 98 415 415 Mean 1.36% 1.58% 1.87% 1.45% 1.84% St. Dev. 8.34% 5.60% 13.55% 5.99% 5.52% Skewness -0.27 0.56 1.02 -0.77 -0.55 Excess Kurtosis -1.74 1.40 0.81 1.32 -0.09 Median 1.48% 1.34% 0.81% 1.57% 1.92% Min -30.51% -17.57% -30.87% -29.24% -26.92% Max 27.12% 24.05% 62.72% 26.35% 21.02% Sharp 0.33 0.72 0.36 0.41 0.69
Table 8. Distributional characteristics of monthly returns of emerging markets, denominated in US$
Czech Republic Hungary Poland Russia Turkey Slovakia Bulgaria Croatia Estonia Kazakhstan Lithuania
26
Table 8. continued
Romania Serbia Slovenia Ukraine
27
Appendix 4:
In this Appendix we look deeper into the Fama and French (1996) three-factor model as explanatory factor for excess returns in emerging markets. In previous chapters we already looked into this model but now we focuses on time variation. We used the same regression variables and dataset as was explained in the data section and methodology section. We show the regressions based on 10 year periods and rolling over time using a 5 year time period. We saw the R-squared of the model become higher over time.
Table 9. Fama and French three-factor model. We show the three-factor and single factor models over
the entire period and over a time period of ten years. For the purpose of explaining the excess returns in emerging markets.
28
Fig 11.The Fama and French model. We show in this figure the Beta(developed market excess
return), HML and SMB as explanatory variables and excess return in emerging markets as a dependent variable. In Panel A and B the constant and beta of the regression are shown. Panel C shows the R-squared of the calculated model.
Panel A: Constant (Alpha) Panel B: Fama and French variables
29
Appendix 5:
In this Appendix we look deeper into the leading indicator as explanatory factor for excess returns in emerging markets. We looked into the time variation of the relation by using a regression over time. In calculating the regressions we used a 5 years period, rolling over the years 1976 till 2010. We used the following regression over time:
Where is the term for the leading indicator.
Fig 12. Leading indicator regression. We show in this figure the leading indicator as
explanatory variables and excess return in emerging markets as depended variable. In Panel A and B the constant and beta of the regression are shown. In Panel C the R-squared of the calculated model is shown.
Panel A: Constant (Alpha) Panel B: Leading indicator beta
30
Fig 13. Leading indicator correlation. We show in this figure the five years rolling correlation
31
References:
Anderson J., 2009, “Poised to Perform”, Far Eastern Economic review, Vol. 172, No. 7, p8-13.
Arora, R.K., Jain P.K., Das, H., 2009, “Behavior of stock returns in selected emerging markets.”, Journal of Financial Management and Analysis, Vol. 22, No. 2, p13-25.
Bekaert, G., 1995, “Market integration and investment barriers in emerging equity markets”, World Bank Economic Review, Vol. 9, p75-107.
Calvo, G., Leiderman, L., and Reinhart, C., 1993, “Capital inflows and the real exchange rate appreciation in Latin America: The role of external factors.”, IMF Staff Papers, No. 40, p108-150. Dimson, E., Marsh, P., Staunton, M., 2003, “Global evidence on the equity risk premium.” Journal of Applied Corporate Finance, Vol. 15, p27-38.
Drew, M. E., Veeraraghavan, M., 2003, “Beta, Firm Size, Book-to-Market Equity and Stock Returns: Further Evidence from Emerging Markets”, Journal of the Asia Pacific Economy, Vol. 8, No. 3, p354-79.
Fama, E. F. and French, K. R., 1996, ‘Multifactor explanations of asset pricing anomalies’, Journal of Finance, Vol. 51 p55–84.
Goetzmann, W., Jorion, P., 1999, “Global stock markets in the twentieth century.” Journal of Finance, Vol. 54, p953–980.
Grootveld, H., Hallerbach, W., 1999. "Variance vs downside risk: Is there really that much difference?", European Journal of Operational Research, Vol. 114, No. 2, p304-319.
International Monetary Fund, 2010, “Global Financial Stability Report”, World economic and Financial Surveys, April, Ch. 4, p119-149.
Johnson, R., Soenen, L., 2009, “Commodity prices and stock market behavior in South American countries in the short run.”, Emerging markets, Finance & Trade, Vol 45, No 4, p69-82.
Khandani, A. E., Lo, A. W., 2009, “Illiquidity Premia in Asset Returns: An Empirical Analysis of Hedge Funds, Mutual Funds, and U.S. Equity Portfolios.”, Working paper.
Kohers, G., Kohers N., Kohers, T., 2006, "The risk and return characteristics of developed and emerging stock markets: the recent evidence", Applied Economics Letters, Vol. 13, No 11, p737-743. Ljung, G., Box, G., 1978, “On a measure of lack of fit in time series models.”, Biometrika, Vol. 66, p67-72.
Neumeyer, Pablo A. and Fabrizio Perri, 2004,”Business Cycles in Emerging Markets: The Role of Inter est Rates.” Working Paper.
Rouwenhorst, K. G.,1999, “Local Return Factors and Turnover in Emerging Stock Markets”, Journal of Finance, Vol. 54, No. 4, p1439-1464.
Salomons, R., Grootveld, H., 2003, “The Equity risk premium: Emerging versus Developed Markets.”, Emerging Markets Review, Vol. 4, No. 2, p121-144.
Shackman, J.D., 2006, “The equity premium and market integration: Evidence from international data.”, Journal of International Financial Markets, Institutions and Money, Vol. 16, p155-179. Sharpe, W., 1966, “Mutual fund performance”, Journal of Business, Vol. 39, p119-138.
32
