Management Practices and the Performance of Mutual Fund in the Caribbean

Transcription

1 Management Practices and the Performance of Mutual Fund in the Caribbean By Winston Moore Department of Economics The University of the West Indies, Cave Hill Campus Barbados June 2010

2 Table of Contents Acknowledgement... iv Executive Summary... v 1. Introduction Review of Previous Literature Classification of Mutual Funds Does Management Style Impact on Mutual Fund Performance Persistence in Mutual Fund Returns Mutual Fund Returns Are Mutual Fund Returns Persistent? Determinants of Excess Returns Evaluating the Importance of Various Characteristics of Mutual Fund Managers Evidence of Stock Picking Evidence of Market Timing Conclusions and Policy Recommendations References Figures Figure 3.1: Monthly Mutual Fund Returns Tables Table 3.1: Descriptive Statistics for Monthly Mutual Fund Returns Table 3.2: Descriptive Statistics Table 3.3: Persistence in Abnormal Returns ii

3 Table 4.2: Estimated Models of Stock Picking Table 4.3: Estimated Aggregate Models of Market Timing Table 4.4: Estimated Disaggregated Models of Market Timing iii

4 Acknowledgement The author would like to thank First Caribbean International Bank for generously providing a grant to finance this project. The project also benefited from the research assistance supplied by Stacia Howard. iv

5 Executive Summary The Caribbean mutual fund industry is one of the fastest growing segments of the financial system. In spite of the industry s growing importance to the regional financial system, the management practices of these investment vehicles are largely unknown. The way in which fund managers make their decisions is important since these choices can have effects on investor returns as well as overall system risk. The present study therefore provides an assessment of the extent to which excess or abnormal returns, if they exist, are due to management abilities. The efficient market hypothesis predicts that it is impossible to beat the market, as stock prices incorporate all relevant information. This suggests that mutual fund returns above the market return should be transitory rather than persistent. One way to evaluate whether some mutual funds have some advantage over the others, i.e. managerial ability or stock-picking talent, is to test for persistence in returns. v

6 Persistence in the context of this study simply means the tendency of mutual funds reporting abnormal returns in year to also report abnormal returns in year. A panel autoregression model was employed to measure the persistence in mutual fund returns. In agreement with a priori expectations, mutual funds that report positive returns in the previous year tend to obtain returns that are 12.1 percent higher in the current period. There are two potential sources of these persistent returns: (1) stock selection ability, and; (2) market timing. The estimated models of stock picking ability suggest that most of the abnormal returns obtained by mutual funds can be explained by excess market returns and momentum, i.e. investing in so-called hot stocks. The other two factors, size and book-to-market are insignificant at the aggregate level. The intercept variable, which provides a measure of the stock picking talent of mutual fund managers, was negative and statistically significant, suggesting fund managers do not seem to have any superior informational advantage in terms of picking stocks. In terms of the results for market timing, the Treynor and Mazuy (1966) model indicates that the mutual fund industry tends to time upswings in the market and therefore shift their portfolio to hold more domestic equities. However, the Henriksson and Merton (1991) market-timing coefficient is not statistically different from zero. While these results may seem contradictory, they could be reconciled by noting that Treynor and Mazuy could be capturing shifts in mutual fund portfolios to equities when large positive market swings are likely to occur, while the Henriksson and Merton model capture any positive swing in the market. If this is the case, it vi

7 therefore suggest that mutual fund managers in Barbados seem to be able to time large upswings in the market, but are not quite able to anticipate smaller upticks. The results in this study therefore provide some basic rules-of-thumb that should be of use to individuals interested in investing in mutual funds in Barbados: (1) funds with abnormal returns in the current year are more likely to report higher-than-average returns the following year; (2) avoid funds with persistently low returns, and; (3) do not expect abnormal returns to last forever. The findings reported in this study also imply that mutual funds provide an opportunity for a wide variety of investors to benefit from upswings in the market, without as large a degree in volatility. vii

8 1. Introduction The Caribbean mutual fund industry is one of the fastest growing segments of the financial system. Since the first mutual funds were created in Jamaica, Trinidad and Tobago and Barbados in the 1970s and 1980s, the number of funds available across the region has risen quite rapidly. At the end of 2006, there were about 63 regionally based mutual funds, with 34 registered in Trinidad and Tobago, 18 in Jamaica and 11 in Barbados. The number of mutual fund offerings has grown in tandem with the consumers demand for this product. While the industry only had one entity offering pooled investment accounts in 1987, by the end of 2007 there were 11 funds listed on the Barbados Stock Exchange. In terms of market value, Alleyne and Moore (2006), using data collected from published financial statements, report that the industry has jumped from less than one percent of gross domestic product (GDP) in 1997 to just over 5 percent of GDP by

9 Although the industry has grown in importance to the regional financial system, the investment and management practices of these investment vehicles are largely unknown. The way in which fund managers make their investment decisions is important since these choices can have effects on overall system risk as well as investor returns. The previous empirical literature has assessed the management practices of mutual funds by evaluating the extent to which the stock picking ability and market timing of fund managers are able to earn their investors abnormal returns. In the US, Treynor and Mazuy (1966) as well as Henriksson (1984) both find support of market timing in the medium-term, i.e years. However, Grinblatt et al. (1995) and Carhart (1997) both uncover little or no confirmation of abnormal mutual fund returns due to superior stock selection ability. The mutual fund industry in the Caribbean is a largely under-researched area. The most recent study done on the Barbadian industry is that by Alleyne and Moore (2006) which estimates a model of the demand for mutual funds in Barbados between 1987 and The results obtained suggest that investment flows into mutual funds are not largely driven by economic fundamentals, but instead national income and the previous levels of investment in mutual funds. Alleyne and Moore argue that this result was due to the tax incentive structure in the island that biases investment towards mutual funds. The present report contributes to the literature by evaluating the extent to which excess or abnormal returns, if they exist, are due to management abilities. The two areas evaluated are stock market timing and stock-picking ability. The findings from the report should indicate whether or not mutual fund managers have some informational advantage over other investors 2

10 that therefore allows them to obtain higher than normal returns. It could also provide potential investors some indication of the funds that have had historically superior returns. The remainder of the report is structured as follows. After the introduction, Section 2 summarises the literature evaluating the relationship between stock market returns and managerial practices. If mutual fund managers have some informational advantage then returns should be persistent, and Section 3 of the study evaluates the persistence of returns in Barbados. Section 4 of the paper attempts to evaluate the source of this persistence in mutual funds, while Section 5 offers a summary of the main findings of the study as well as some policy recommendations. 3

11 2. Review of Previous Literature 2.1 Classification of Mutual Funds Mutual funds tend to be grouped together based on the type of financial assets they hold or the styles of their manager. Using the financial asset classification, one can have money market, bond, equity, dividend, real estate or a balanced fund. Most of these classifications are apparent, however, the idea of a money market or balanced fund needs further explanation. Money market funds generally invest in money market securities (e.g. government securities, certificates of deposit, commercial paper of companies, etc.). As a result, dividends tend to be closely related to short-term interest rates. Balanced funds, on the other hand, attempts to strike a balance between the other types of funds and therefore may include stock, bond as well as real estate components. Rather than classifying funds based on their asset holdings, it is also quite common to use the investment style of the fund manager to group these funds. These styles can be grouped together into three main categories: (1) value; (2) growth, and; (3) blend. Value funds tend to invest in established companies and rely on the dividends received to generate value for their investors. 4

12 Growth funds, in contrast, buy shares in companies they expect to experience significant growth in earnings in the future and therefore result in significant gains in terms of stock market value. Balanced or blended funds purport to strike a balance between the two approaches described above. Brown and Goetzmann (1997), however, reject these standard industrial classifications arguing that while these classifications are popular they do not provide any actual indication of the strategies of investment managers nor do they help to explain differences in future returns. The authors instead group funds together using a switching regression and monthly returns on mutual funds. The approach identifies a few major types of funds: equity income, growth and income and growth. 2.2 Does Management Style Impact on Mutual Fund Performance Given the differences in the various management styles that exist between mutual funds, many authors have sought to assess whether these approaches lead to any differences in mutual fund performance. One of earliest papers in the category is that by Grinblatt and Titman (1992). The authors use a database containing 279 mutual funds over the period 1974 to 1984 to assess whether there is any persistence in the performance of these funds. According to the efficient market hypothesis it is impossible to beat the market, as stock prices incorporate all relevant information. This suggests that equity fund returns above that of the market return should be transitory rather than persistent. The funds were grouped into one of twelve portfolio strategies and the risk adjusted performance measure is employed to compare returns. Grinblatt and Titman 5

13 find that there is positive persistence in mutual fund performance, which cannot be explained by firm size, dividend yields, past returns, skewness, interest rate sensitivity or the CAPM beta. One of the shortcomings of this early study by Grinblatt and Titman (1992) is that it did not account for survivorship bias, i.e. only those funds that consistently report returns above average tend to survive. This characteristic could therefore explain the persistence of returns obtained. Brown and Goetzmann (1995) attempt to control for this bias by using a model that explicitly accounts for sample selection bias. The study still reports evidence of persistence, particularly across funds that followed a similar strategy or hot hands /momentum. Similar results are described by Bollen and Busse (2005), Horst and Verbeek (2000), Golect (1996), Grinblatt et al. (1995) Goetzmann and Ibbotson (1994) for developed markets while Kaminsky et al. (2004) finds evidence of momentum trading for mutual fund managers in emerging markets. In a similar vein, Chevalier and Ellison (1999a) assess the extent to which these dissimilarities may be due to variations in the ability, knowledge or effort of the managers that are in charge of these funds. Three variables were assessed: the manager s age, the quality of the undergraduate institution attended and whether the manager has an MBA. Chevalier and Ellison report that managers that attended more selective undergraduate institutions tend to outperform their counterparts. These results could suggest that there are direct benefits in terms of returns from having a manager with a better education. There was also an (inverse albeit weak) relationship between mutual fund performance and age. In a follow-up study, Chevalier and Ellison (1999b) find that for young managers the probability of termination increases steeply with performance when returns are negative, but is insignificant at positive excess returns levels. As a result, young managers may have an incentive to avoid unsystematic risk when selecting their portfolios. 6

14 Golec (1996), however, finds that it is not the age of the manager but rather the length of his/her tenure at their fund. Gottesman and Morey (2006) also find fund managers holding MBAs from one of the top 30 business schools are more likely to report statistically superior performance. Elton et al. (2003) argue that superior performance from fund managers can be extracted through the use of incentive fees. Incentive fees align manager interest with investor interests. As a result, both the investors in the fund as well as the fund manager do better when the fund reports excess returns. Using a database containing 108 funds between 1990 and 1999, Elton et al. (2003) report no significant difference in the average return of funds where the manager s compensation is linked to the performance of the mutual fund. Nevertheless, investors in mutual funds with incentive fees do benefit from attracting managers with better stock selection ability as well as lower expense ratios. Similar benefits were also reported for mutual funds owned by the portfolio manager by Khorana et al. (2007). Prather and Middleton (2002) evaluate the relative performance of mutual funds managed by individuals or teams. The authors use monthly continuously compounded risk-adjusted net returns of 162 open-end mutual funds over a 13-year period. The mutual funds are classified into eight investment categories based on management style and asset holdings as well as rather or not the funds were managed by individuals or teams. The results, however suggest that there was no substantial difference in the results of mutual funds managed by a team or an individual. Prather and Middleton note that this finding agrees with classical utility theory which states that differing alternatives to the same problem should result in the same choice whether or not that decision is taken by an individual, group or company. Atkinson et al. (2003) also note that there is no 7

15 difference in performance and investment behaviour of funds managed by male or female fixedincome mutual fund managers. 8

16 3. Persistence in Mutual Fund Returns If markets are efficient, they should incorporate all relevant information and it should therefore be impossible to beat the market. Above average mutual fund returns should therefore be transitory rather than persistent. One way to evaluate whether some mutual funds have some advantage over the others, i.e. managerial ability or stock-picking talent, can be evaluated by testing for persistence in mutual fund returns. 3.1 Mutual Fund Returns The study employs monthly observations on Barbadian mutual funds over the period January 2003 to April The data is free from survivor bias, since none of the mutual funds exited over this period. Observations on net asset values for the mutual funds are obtained from the Daily Trading Reports published by the Barbados Stock Exchange and available at 9

17 The monthly returns ( on each fund are calculated as: (3.1) where is the net asset value of the fund in month. Figure 1 provides a plot of these monthly returns along with the return for the market. There was no clear trend in returns over the sample period, with most funds fluctuating around the mean zero level. Monthly returns also appear to be highly correlated with the monthly return for the Barbados Stock Exchange Domestic Index. The volatility for most funds are, however, lower than that for the overall market index. Figure 3.1: Monthly Mutual Fund Returns BNB Capital Growth BNB Incom e BNB Property Clico Balanced Fortress Caribbean Growth Fortress High Interest Acc. Fortress High Interest Dist. Royal Fidelity Select Balanced Sagicor Global Balanced Sagicor Select Growth Sagicor Preferred Income RNAV RMKT 10

18 Table 3.1 present the descriptive statistics for the monthly mutual fund returns in Barbados between 2003 and Average monthly returns over the period was 0.3 percent, with Fortress Caribbean Growth reporting the highest mean monthly returns over the entire sample period of 0.7 percent. Other funds reporting above average returns were BNB Property (0.4 percent), Fortress High Interest (0.5 percent), Royal Fidelity Select Balanced (0.5 percent) and Sagicor Global Balanced (0.4 percent). Only one fund, Sagicor Select Growth, had mean negative returns over the sample period. Table 3.1: Descriptive Statistics for Monthly Mutual Fund Returns Mean Max Min. Std. Dev. Skew. Kurt. Obs. BNB Capital Growth Fund BNB Income Fund BNB Property Fund CLICO Balanced Fund Fortress Caribbean Growth Fund Fortress High Interest Fund Acc Fortress High Interest Fund Dist Royal Fidelity Select Balanced Fund Sagicor Global Balanced Fund Sagicor Select Growth Fund Sagicor Preferred Income Fund Average Growth funds had the highest average rates of volatility, with the Sagicor Select Growth, Fortress Caribbean Growth and the BNB Capital Growth. Most returns are negatively skewed indicating that observations are clustered in the negative region. The measure of excess kurtosis for all the mutual funds deviate significantly from that expected from returns drawn from a normal 11

19 distribution, i.e. 3. In particular, BNB Property, Fortress Caribbean Growth, BNB Income Fund, Fortress High Interest Fund and Sagicor Select Growth all had measured excess kurtosis significantly above 3. The non-normality is confirmed by the significance of the Jarque-Bera statistic. 3.1 Are Mutual Fund Returns Persistent? Persistence in the context of this study simply means the tendency of mutual funds reporting abnormal returns in year to also report abnormal returns in year. If this is the case, a panel autoregression model can be employed to measure the persistence in mutual fund returns. The autoregression takes the form: (3.2) where is an error term that is assumed to have normal properties. If mutual fund returns are persistent then over the sample period, indicating that positive returns last year should be followed by positive returns in the current year. The coefficient estimates for the equation above is obtained using a pooled model with white robust cross-sectional standard errors. 1 The results from estimating Equation (3.2) are provided in Table 3.2. The model is able to explain a relatively large proportion of the fluctuation in mutual fund returns, about one-third. In agreement with a priori expectations mutual fund 1 A likelihood ratio test for redundant fixed effects had a test statistic of and a p-value of

20 returns are persistent, with mutual funds reporting positive returns in the current year expected to obtain a 12.1 percent rise in returns the following year. Table 3.2: Descriptive Statistics Dependent Variable = Returns constant (14.258)** (1.962)** R-squared s.e. of regression F-statistic [0.000] Notes: (1) t-statistics are provided in parentheses below coefficients. (2) p-values are given in square brackets. There are two main drawbacks of the autoregression approach. First, the equation evaluates the persistence in returns rather than abnormal returns. It is therefore possible that the high degree of persistence obtained in Table 3.2 could be due to under-performing mutual funds. Second, as noted by Grinblatt and Titman (1992) the t-statistic reported in the Table 3.2 does not have a true t-distribution as some funds may have similar portfolios and therefore the residuals could be highly correlated. Following Grinblatt and Titman (1992) persistence can alternatively be evaluated using a three-step procedure. In the first stage, the sample is split into two three-year sub-periods: and Second, the abnormal returns for each fund are computed. The abnormal returns for a particular fund are defined as the excess of the average abnormal return of all funds over the 3-year period. Finally, the multiple r-squared of the abnormal returns in the first three-years and the second three-years is calculated. A statistically significant correlation between returns in the two periods would suggest that returns are highly 13

21 related to those obtained in the prior period and could indicate some difference in managerial ability or stock picking talent. Table 3.3 calculates the abnormal returns for mutual funds using the 3-month t-bill rate. The averages of these abnormal returns for the two consecutive 3-year periods are provided in Table 3.3. The squared Pearson correlation coefficient imply that abnormal returns in the two periods are correlated, i.e. mutual funds that performed well in the period also tended to perform above par in the period. Indeed the slope coefficient suggest that mutual funds reporting abnormal returns in the period are likely to obtain returns that are 4.5 percent greater in the period. The coefficient is significant at the 10 percent level of testing. Table 3.3: Persistence in Abnormal Returns Abnormal Returns above T-bill Rate BNB Capital Growth Fund BNB Income Fund BNB Property Fund CLICO Balanced Fund Fortress Caribbean Growth Fund Fortress High Interest Fund Acc Fortress High Interest Fund Dist Royal Fidelity Select Balanced Fund Sagicor Global Balanced Fund R-squared Sagicor Select Growth Fund Slope Coefficient Sagicor Preferred Income Fund p-value

22 4. Determinants of Excess Returns 4.1 Evaluating the Importance of Various Characteristics of Mutual Fund Managers While the Grinblatt and Titman (1992) approach can provide evidence of differences in the performances of the managers of mutual funds, it cannot identify the source of these differentials. One potential source of persistent abnormal returns is stock selection ability. To evaluate whether or not a particular fund s return is due to stock selection ability Carhart (1997) uses the following four-factor model: (5.1) where is the excess return of the mutual fund and are the returns of four factors. These four factors are the excess return of the market portfolio, size, book-to-market and momentum factors. 15

23 Following Fama and French (1993), the size factor is estimated as the difference in monthly returns in small-stock portfolios and big-stock portfolios, while book-to-market factors are computed by the differences in the average returns of firms with high book-to-market equity ratios and low book-to-market equity values. Following Carhart (1997), momentum is defined as the difference in returns of firms with above median returns lagged one month minus the average returns of firms with below median returns lagged one month. The intercept from Equation 5.1 is then employed to measure abnormal returns from picking stocks that outperform a risk-adjusted benchmark. The model is then estimated over 3-year horizons and the intercept used to rank each fund. Another potential difference in managerial practices could arise from market timing, i.e. correctly forecasting the relative returns of broad asset classes. Treynor and Mazuy (1966) suggest that market timing can be detecting using the following regression: (5.2) where is a measure of market timing ability. This captures the idea that if a fund manager increases a portfolio s exposure to equities in advance of positive excess market returns, the fund s return should be positive and statistically significant. Similarly, Henriksson and Merton (1981) captures this non-linear relationship between the fund s excess return and the market excess return by using an indicator variable,, that takes a value of 1 if the market s excess return is above zero and zero otherwise. The estimated regression is of the following form: (5.3) 16

24 similar to Equation (5.2) is expected to be positive and statistically significant as this would indicate that the fund manager is increasing the fund s exposure to equity whenever there is a market rally. 4.2 Evidence of Stock Picking Table 4.2 provides two models of the stock picking abilities of mutual fund managers. The first model assesses the ability of the industry in general. The results suggest that most of the fluctuation in mutual fund returns can be explained by excess market returns and momentum. This indicates that mutual funds tend to invest in so-called hot stocks and therefore benefit from the momentum in returns. The other two factors, size and book-to-market were, however, insignificant at the aggregate level. The intercept variable, which provides a measure of the picking talent of mutual fund managers, was negative and statistically significant. Similar to Carhart (1997) most of the abnormal returns or persistence in mutual fund returns in Barbados can be accounted for by market returns and investing in hot stocks. This therefore suggests that mutual fund managers do not seem to have any superior informational advantage in terms in picking stocks. It is possible, however, that while the industry may not benefit from any advantages in terms of picking stocks, individual funds could be experiencing some advantages from the stock picking talent of their managers. Table 4.2 therefore also provides the coefficient estimates from a firmspecific model of stock picking. When these firm-specific factors are considered all four factors proved to be statistically significant at normal levels of testing. Nevertheless, the coefficient estimates for the intercept for each mutual fund while statistically significant was negative. 17

25 Similar to the aggregate market model therefore, there is no evidence of superior stock-picking talent. Table 4.2: Estimated Models of Stock Picking Industry Firm-Specific ( )** (19.951)** (4.496)** (-0.226) (2.529)** (-0.949) (-3.360)** (2.361)** (1.711)* BNB Capital Growth Fund ( )** BNB Income Fund ( )** BNB Property Fund ( )** CLICO Balanced Fund ( )** Fortress Caribbean Growth Fund ( )** Fortress High Interest Fund Acc ( )** Fortress High Interest Fund Dist ( )** Royal Fidelity Select Balanced Fund ( )** Sagicor Global Balanced Fund ( )** Sagicor Select Growth Fund ( )** Sagicor Preferred Income Fund ( )** R-squared S.e regression F-Statistic

26 [0.000] [0.000] Notes: (1) t-statistics are provided in parentheses below coefficients. (2) p-values are given in square brackets. (3) ** and * indicates statistical significance at the 5 and 10 percent levels of testing, respectively. 19

27 4.3 Evidence of Market Timing Table 4.3 supplies the results from estimating the models of market timing outlined in Section 4.1. Both the Treynor and Mazuy (1966) and Henriksson and Merton (1981) models are given for comparison purposes. In this initial table, no attempt is made to differentiate between the performances of individual mutual funds. The results therefore apply to the industry in general. The coefficient on the market portfolio is positive and statistically significant, in both the Treynor and Mazuy (1966) and Henrisksson and Merton (1981) models. These results confirm the initial analysis of data conducted in Section 3 of the study and indicates that excess returns on mutual funds tend to be highly correlated with excess market returns. In terms of the coefficient of interest,, in the Treynor and Mazuy model the coefficient is positive and statistically significant at normal levels of testing and therefore indicates that the mutual fund industry tends to time upswings in the market and therefore shift their portfolio to hold a larger proportion of domestic equities. In the case of the Henriksson and Merton model, however, the coefficient is not statistically different from zero. While these results may seem contradictory, they could be reconciled by noting that Treynor and Mazuy could be capturing shifts in mutual fund portfolios to equities when large positive market swings are likely to occur while the Henriksson and Merton model capture any positive swing in the market. If this is the case, it therefore suggest that mutual fund managers in Barbados seem to be able to time large upswings in the market, but are not quite able to anticipate smaller upticks. 20

28 Table 4.3: Estimated Aggregate Models of Market Timing Treynor and Mazuy (1966) Henriksson and Merton Model (1981) ( )** (-4.983)** (20.615)** (3.819)** (2.838)** (-0.085) R-squared S.e regression F-Statistic [0.000] [0.000] Notes: (1) t-statistics are provided in parentheses below coefficients. (2) p-values are given in square brackets. (3) ** and * indicates statistical significance at the 5 and 10 percent levels of testing, respectively. The aggregate market timing models are provided in Table 4.3. While these models are able to evaluate whether the mutual fund industry, on average, is able to anticipate positive swings in the market, they do not provide a comparison of the market timing capabilities of individual mutual funds. To provide this assessment is interacted with dummy variables that take a value of 1 for mutual fund and zero otherwise. These coefficients therefore provide mutual fund specific estimates of market timing. The estimated disaggregated models of market timing are given in Table 4.4. The intercept and coefficient on market returns are quite similar to those obtained earlier for both models. In the case of the Treynor and Mazuy (1966) model 6 out of the 11 funds considered had positive and statistically significant market timing coefficients. These 6 funds had various investment objectives ranging from income to property. In contrast, the market timing coefficients from the 21

29 Henriksson and Merton (1981) model were statistically insignificant from zero for most funds. However, 2 funds had negative market timing coefficients indicating that returns tend to be inversely related to upswings on the domestic market Table 4.4: Estimated Disaggregated Models of Market Timing Treynor and Mazuy Henriksson and Merton (1966) Model (1981) ( )** (-4.941)** (20.101)** (3.840)** BNB Capital Growth Fund BNB Income Fund BNB Property Fund CLICO Balanced Fund Fortress Caribbean Growth Fund Fortress High Interest Fund Acc Fortress High Interest Fund Dist Royal Fidelity Select Balanced Fund Sagicor Global Balanced Fund Sagicor Select Growth Fund Sagicor Preferred Income Fund (-0.234) (1.673)* (3.045)** (0.575) (2.163)** (1.843)* (0.263) (2.062)** (1.848)* (0.852) (0.153) (-0.030) (-0.311) (-0.576) (0.065) (1.014) (-0.869) (-1.219) (0.522) (0.274) (5.827)** (-4.164)** R-squared S.e regression F-Statistic [0.000] [0.000] Notes: (1) t-statistics are provided in parentheses below coefficients. (2) p-values are given in square brackets. (3) ** and * indicates statistical significance at the 5 and 10 percent levels of testing, respectively. 22

30 23

31 5. Conclusions and Policy Recommendations Mutual funds are usually classified either by the type of financial assets they hold or the styles of their manager. Brown and Goetzmann (1997), however, argue that these standard industry-wide classifications do not provide an adequate indication of management strategies or explain future returns. In the Caribbean, various categories of mutual funds are available, for example income, growth, property, etc. To date, however, there has been no attempt to evaluate what impact management strategies have on the performance and returns of mutual funds in the region. This study attempts to fill this gap in the literature by assessing the extent to which the stock picking and market timing ability of managers can lead to abnormal returns for mutual fund investors. The study uses monthly observations on 11 mutual funds in Barbados between January 2003 and April To evaluate whether some mutual funds tend to outperform their counterparts over time a panel autoregression was estimated. The results provide some initial evidence of persistence with the coefficient on the lagged return variable being positive and statistically significant. This finding was also collaborated by using the Grinblatt and Titman (1992) which is more robust to correlated errors resulting from mutual funds holding similar portfolios. 24

32 To assess the determinants of this persistence in mutual fund returns, the study ascertain the validity of two hypotheses: (1) superior stock selection talent, or; (2) market timing. The estimated model results suggest that mutual fund managers do not seem to have any superior informational advantage that would allow them to select better performing shares. Instead, most of the variation in mutual fund returns can be explained by investing in hot stocks, large cap shares and firms with relatively high book-to-market equity valuations. There was, however, some evidence of market timing both at an aggregate industry level and for individual mutual funds. Indeed, 6 out of the 11 funds considered tended to shift their portfolios to hold more equity when there were large upswings in the market. The study therefore provide some basic rules-of-thumb that should be of use to individuals interested in investing in mutual funds in Barbados: (1) funds with relatively high returns are more likely to report higher-than-average returns the following year; (2) avoid funds with persistently low returns, and; (3) do not expect abnormal returns to last forever. The results supplied in this paper suggest that mutual funds offer investors the opportunity to benefit from upswings in the market, without as large a degree in volatility. 25

33 References Alleyne, K. and Moore, W. (2006) The Economic Effects of Tax Incentives on the Mutual Fund Industry in Barbados, Journal of Eastern Caribbean Studies, Vol. 31, pp Arellano, M. and Bond, S. (1991) Some tests of specification for panel data: Monte Carlo evidence and an application to employment equation. Review of Economic Studies, Vol. 58, pp

34 Atkinson, S.M., Baird, S.B. and Frye, M.B. (2003) Do Female Mutual Fund Managers Manage Differently, Journal of Financial Research, Vol. 26, pp Bollen, N.P.B. and Busse, J.A. (2005) Short-term Persistence in Mutual Fund Performance, Review of Financial Studies, Vol. 18, pp Brown, S.J. and Goetzmann, W.N. (1997) Mutual Fund Styles, Journal of Financial Economics, Vol. 43, pp Brown, S.J. and Goetzmann, W.N. (1995) Performance Persistence, Journal of Finance, Vol. 50, pp Carhart, M.M. (1997) On Persistence in Mutual Fund Performance, Journal of Finance, Vol. 52, pp Chan, L.K.C., Chen, H-L. and Lakonishok, J. (2002) On Mutual Fund Investment Styles, Review of Financial Studies, Vol. 15 (5), pp Chevalier, J. and Ellison, G. (1999a) Are Some Mutual Fund Mangers Better than Others? Cross-Sectional Patterns in Behaviour and Performance, Journal of Finance, Vol. 54, pp Chevalier, J. and Ellison, G. (1999b) Career Concerns of Mutual Fund Managers, Quarterly Journal of Economics, Vol. 114, pp Elton, E.J., Gruber, M.J. and Blake, C.R. (2003) Incentive Fees and Mutual Funds, Journal of Finance, Vol. 58, pp Goetzmann, W.N. and Ibbotson, R.G. (1994) Do Winners Repeat? Patterns in Mutual Fund Performance, Journal of Portfolio Management, Vol. 20, pp Golec, J.H. (1996) The Effects of Mutual Fund Managers Characteristics on their Portfolio Performance, Risk and Fees, Financial Services Review, Vol. 5, pp

35 Gottesman, A.A. and Morey, M.R. (2006) Manager Education and Mutual Fund Performance, Journal of Empirical Finance, Vol. 13, pp Grinblatt, M., Titman, S. and Wermers, R. (1995) Momentum Investment Strategies, Portfolio Performance, and Herding: A Study of Mutual Fund Behaviour, American Economic Review, Vol. 85, pp Grinblatt, M. and Titman, S. (1992) The Persistence of Mutual Fund Performance, Journal of Finance, Vol. 47, pp Henriksson, R. (1984) Market Timing and Mutual Fund Performance: An Emprical Investigation, Journal of Finance, Vol. 51, pp Henriksson, R. and Merton, R. (1981) On Market Timing and Investment Performance. II. Statistical Procedures for Evaluating Forecasting Skills, Journal of Business, Vol. 54, pp Horst, J. and Verbeek, M. (2000) Estimating Short-Run Persistence in Mutual Fund Performance, Review of Economics and Statistics, Vol. 82, pp Kaminsky, G., Lyons, R.K. and Schmukler, S.L. (2004) Managers, Investors, and Crises: Mutual Fund Strategies in Emerging Markets, Journal of International Economics, Vol. 64, pp Khorana, A., Servaes, H. and Wedge, L. (2007) Portfolio Manager Ownership and Fund Performance, Journal of Financial Economics, Vol. 85, pp Prather, L.J. and Middleton, K.L. (2002) Are N+1 Heads Better than One? The Case of Mutual Fund Managers, Journal of Economic Behaviour and Organization, Vol. 47, pp Treynor, J. and Mazuy, K. (1966) Can Mutual Funds Outguess the Market? Harvard Business Review, Vol. 44, pp

36 29