Selection ability and market timing skills of mutual fund and unit trust managers in a developing economy: evidence from Ghana

Abstract

Purpose

The study assess the selection ability and market timing skills of mutual fund and unit trust managers in Ghana.

Design/methodology/approach

The study uses an improved survivorship bias-free dataset of yearly after-fee returns of all mutual funds and unit trusts operating in Ghana from January 2011 to December 2019, cumulating in nine years of quantitative fund data. The authors assess Mutual funds and Unit trusts that ever existed, “alive” or “dead,” over the sample period in the study. The authors construct factor loadings to enable the application of multifactor models in the analysis. The authors apply the unconditional versions of the Jensen alpha, Fama-French three-factor, and Carhart four-factor models to determine the selection ability and market timing skills of 32 mutual funds and 17 unit trusts. The authors deploy HAC-consistent robust standard errors to the OLS estimations to subdue the effect of heterogeneity and autocorrelation.

Findings

The results indicate that, on average, mutual funds and unit trust managers possess market timing skills but no selection ability. When the results are decomposed into fund types, fixed-income and balanced mutual fund managers possess selection ability and market timing skills.

Originality/value

To the authors' best knowledge, this study is the earliest to examine the selection ability and market timing skills of both mutual fund and unit trust managers in Sub-Saharan Africa (SSA). It is also the earliest to construct factor loadings for the Ghana stock market.

Citation

Danquah, R. and Yu, B. (2023), "Selection ability and market timing skills of mutual fund and unit trust managers in a developing economy: evidence from Ghana", Business Analyst Journal, Vol. 44 No. 1, pp. 1-14. https://doi.org/10.1108/BAJ-09-2022-0028

1. Introduction

The importance of mutual funds and unit trusts in financial markets cannot be over-emphasized. Over the last few decades, the net asset values of mutual funds and unit trusts in the USA have been estimated at $23.9tr and $78bn, respectively. Over 60.9mn households and 106.3 mn individuals own mutual funds in the USA, respectively, with $126,700 as households' median mutual fund asset value (Investment Company Institute, 2021). Mutual funds and unit trusts are avenues investors use to save for current and future needs and thus play significant roles in the financial market by mobilizing funds from savers to various economic and production units.

Theoretically, mutual funds and unit trusts net asset values comprise net cash inflows and fund returns. Fund returns directly impact fund cash inflows/outflows and net asset value. Omori and Kitamura (2022) posit that cash inflow from fund investors depends on the fund manager's ability to generate alpha. But generating alpha emanates from the managers' selection ability and market timing skills. Fund selection ability and market timing skills induce cash inflows or outflows into mutual funds and unit trusts. Thus, the study's main objective is to examine mutual fund and unit trust managers' selection ability and market timing skills. Managers' ability to select winner stocks or investments in their portfolios may arise from the sophistication of their research capabilities and the efficiency of the market. With a weak form of market efficiency, and the relatively young nature of the Ghanaian fund industry, asset management firms may not have built enough sophisticated and robust research units to support their portfolio selection capabilities. We, therefore, hypothesize that mutual fund and unit trust managers do not possess selection ability. However, any market which exhibits a weak form of efficiency also provides arbitrage opportunities. We hypothesize that fund managers in Ghana have market timing skills because of arbitrage opportunities and the weak form efficiency of the market.

Since past fund performance predicts future performance Bollen and Busse (2005), Huij and Verbeek (2007), Jegadeesh and Titman (1993), it is helpful to investigate the selection ability and market timing skills of managers. Also, most literature on mutual funds and unit trusts emanates from the Americas, Europe, and Asia. Studies from developing markets in Africa are negligible. It is, therefore, imperative that we undertake this study. Furthermore, developing markets have unique characteristics differentiating them from developed and advanced markets. For instance, information on mutual funds and unit trusts is not widely available in developing markets compared to the Americas and Europe, investor sophistication in developing economies is not as compared to developed and advanced markets, and there is no stiff competition in the fund industry in developing markets. Most fund investors in developing markets subscribe to mutual funds and unit trusts based on word of mouth from family and friends. Therefore, studying funds selection ability and market timing skills are in order.

The current study is different from other studies in several ways. For instance, Tan (2015) evaluate the selection ability and market timing skills of 10 South African equity mutual funds. We do not analyze only equity funds but balanced and fixed-income funds as well. Thobejane, Simo-Kengne, and Muteba Mwamba (2017) examine equity unit trust performance in South Africa. The current study does not only analyze the selection and market timing skills of equity unit trust funds but also balanced and fixed-income unit trusts. Also, IIo, Yinusa, and Elumah (2018) investigated the performance of 6 portfolio classes of 37 mutual funds distributed in Nigeria using Jensen alpha, Sharpe, and Treynor ratios. We examine not only different portfolio classes of mutual funds and unit trusts but also deploy unconditional versions of the Jensen alpha and Fama-French-Carhart (FFC) multifactor models in our estimation. Finally, Musah, Senyo, and Nuhu (2014) examine the selection ability and market timing skills of 8 mutual funds in Ghana from 2007 to 2012. Survivorship bias is possible in the dataset since not all funds existing within the sample period are utilized. This study differs from Musah et al. (2014) because we minimize the potential for survivorship bias by using an improved dataset of all funds that ever existed in Ghana from 2011 to 2019.

The study adds to narrowing the literature gap in the study of mutual fund and unit trust fund managers' selectivity and market timing skills in Africa. Fund investors are rational consumers who want to minimize costs while increasing returns. The study also contributes to investor decision-making as they know which funds have the selection ability and market timing skills to enhance their returns.

We proceed as follows. A review of the literature on fund selection ability and market timing skills is in section 2. Section 3 describes the data and methods employed in the study. We present the empirical results and discussion in section 4, while section 5 concludes.

2. Literature review

Most literature indicates that fund managers do not have selection ability and market timing skills. Fama and French (2010) analyze the performance of actively managed equity mutual funds in the USA from 1984 - 2006, mainly employing their three-factor and Carhart (1997) four-factor models. Fama and French (henceforth, FF) argue cost in expense ratios make mutual fund net returns underperform the CAPM, three-factor and four-factor benchmarks. Therefore, many unskilful managers overshadow managers with the expertise to generate risk-adjusted returns. The authors further suggest stronger managerial skills when returns are measured before factoring in fund fees.

Fund managers of socially responsible (SR) mutual funds have also been assessed for selection ability and market timing skills. Das and Uma Rao (2015) evaluate the selection and market timing skills of SR open-end mutual funds in the USA using monthly data from the Morningstar database from July 2002 to June 2012. The authors engage the Treynor-Mazuy and Henriksson-Merton unconditional factor models in their analysis. They find significant stock selection ability but no market timing skills over the period. The trade-off between selection ability and market timing skills is also evidenced by Thobejane et al. (2017). The study found no evidence of selection ability but strong evidence of market timing skills across all the sample categories. Ferruz, Muñoz, and Vargas (2012) utilized the US net of fees returns dataset of religious equity mutual funds from the Thomson Reuters database from January 1994 to September 2010 to examine their selection and market timing performance. Ferruz et al. (2012) conclude that religious equity mutual funds do not possess selection ability and market timing skills.

Agarwal and Pradhan (2018) investigate the selection ability and market timing skills of 240 open-ended growth equity mutual funds in India from 2006 to 2015. The study controlled survivorship bias using all funds and deployed multifactor versions of Henriksson and Merton (1981) and Treynor and Mazuy (1966) to measure selection and market timing skills. Agarwal and Pradhan (2018) document significant stock selection and market timing skills of Indian equity fund managers. The authors also report the co-existence of market timing skills and selection ability.

Dong and Doukas (2019) analyze the selectivity skill of 2,947 actively managed domestic equities mutual funds from 11 European countries from January 1998 to December 2015. The study obtained data from the Bloomberg database and mimicked Amihud and Goyenko (2013) by deploying the logistically transformed 1-R². Dong and Doukas (2019) evidence fund selection abilities in the European market. The authors find that fund selectivity and fund performance are positively correlated, indicating that managers with greater degrees of selectivity skills produce better risk-adjusted excess returns. And that this correlation is robust after controlling for market sentiment and dispersion. The study also evidence that country-level features moderate the correlation between selectivity and performance of fund managers.

Nikolaos, Alexandros, Stephanos, and Beneki (2020) investigate the selection ability and market timing skills of 17 Greek equity mutual funds from January 2005 to April 2010, utilizing Treynor and Mazuy's unconditional version of CAPM. The authors posit that equity fund managers lack selectivity and market timing skills. However, the lack of selectivity and market timing skills gradually diminishes after the Greek Sovereign debt crisis. Similarly, Tan (2015) finds no selection ability and market timing skills in South African equity mutual funds from 2009 to 2014.

Other studies indicate that managers' performance is purely due to lack but not skill. For instance, Gao, O’Sullivan, and Sherman (2021) use weekly fund returns of a survivorship bias-free dataset of 778 funds from May 2003 to September 2020 and employed the bootstrap methodology. They document the non-existence of “genuinely skilled fund managers”. The authors report that the out-performance of top-ranked funds is purely due to “good luck” but not “good skill” while the under-performance of bottom-ranked funds is attributable to “bad skill”. Stotz (2007) suggests that fund investors should follow their timing strategies since the average German fund manager does not possess market timing ability and portfolio selection skills.

Managers' selection ability and market timing skills induce funds inflow/outflow from mutual funds and unit trusts. Omori and Kitamura (2022) analyze active equity mutual funds in Japan from 2000 – 2019 using monthly data from the Morningstar database. Omori and Kitamura evidence that Japanese active mutual fund investors evaluate funds based on the fund's selection ability which also determines their inflows and outflows.

From the above literature review, it is evident that there are contrasting views on the selection ability and market timing skills of mutual fund and unit trust managers in developed and developing markets. This study is essential to determine the selection ability and market timing skills of fund managers in a developing economy in Sub-Saharan Africa and to know which side of the argument it aligns with.

3. Data and methods

3.1 Data

The study uses an improved survivorship bias-free dataset of yearly after-fee returns of all mutual funds and unit trusts operating in Ghana from January 2011 to December 2019, cumulating in 9 years of quantitative fund data. Brown and Goetzmann (1995) included funds existing at the close of the sample period. We minimize the effect of survivorship bias by using a dataset of funds that are “alive” and “dead” over the sample period, Bollen and Busse (2005). We dropped 18 funds with less than three years of fund data from the study. The study's data is obtained from Ghana Securities and Exchange Commission (SEC), individual asset management companies, Ghana Stock Exchange, Databank research, and an online portal: Annual Reports Ghana (www.annualreportsghana.com). We further group and analyze mutual funds and unit trusts according to fund type (equity, fixed-income, and balanced funds) established by Ghana's SEC. For this study, money market and fixed-income funds are considered fixed-income. We utilize 32 mutual funds and 17 unit trusts in the study. Table 1, lists mutual funds and unit trusts by fund type, and the names of their respective managers used in the study.

Factor loadings must be constructed to deploy the unconditional Jensen alpha, Fama-French (FF) three-factor, and Carhart four-factor models to analyze selection ability and market timing skills. The study mimics Fama-French [1] to construct the model factor loadings (See Table 2) using end-of-year market data from the Ghana stock exchange from 2011 - 2019.

3.1.1 Steps in constructing factor loadings

1. Rank all stocks using their market value of equity. Use that to compute the median value of equity.

2. All stocks with a value below the median value of equity are termed as ‘Small cap,' while all stock values above the median value of equity are termed as ‘Big cap.'

3. Rank all stocks according to the ratio of the book value of equity to the market value of equity.

4. After ranking, we take the 70th and 30th percentiles.

5. Stocks with values above the 70th percentile are classified as ‘Value' stocks.

6. Stocks with values below the 30th percentile are classified as ‘Growth' stocks.

7. Stocks with values less than the 70th but more than 30th percentiles are classified as ‘Neutral' stocks.

8. Take the average returns of both ‘Small value' & ‘Big value' stocks.

9. Take the average returns of stocks of both ‘Small neutral' & ‘Big neutral,' respectively.

10. Take the average returns of stocks of both ‘Small Growth' & ‘Big Growth,' respectively.

11. Calculate the factor loadings SMB, HML, & UMD by using the below methods:

SMB =1/3 (Small Value +Small Neutral +Small Growth)

−1/3 (Big Value +Big Neutral +Big Growth)

HML =1/2 (Small Value +Big Value)

     −1/2 (Small Growth +Big Growth)

UMD =1/2 (highest Small caps return +highest Big caps return)

   – 1/2 (lowest Small caps return +lowest Big caps return)

From Table 2 above, the average excess return in the market is −4.70%, indicating that over the period, the 91-day treasury bill (a proxy for risk-free rate) has outperformed the volume-weighted average GSE-CI index of the stock market. That is, the average return from the stock market over the entire sample period falls short of the average 91-day treasury bill rate. The average SMB of −10.19% shows that low-capitalized stocks outperform higher-capitalized stocks in the market. The 3.7% average HML also indicates that value stocks outclass growth stocks in the market. Average previous high advancing stocks outperform the previous lower advancing stocks by 187.96%, as shown by the UMD over the period.

3.2 Methods

We employ the unconditional Jensen alpha, FF three-factor, and Carhart four-factor models to examine managers' selection ability and market timing skills. To avoid the effect of survivorship bias on our results, we utilize the returns of all funds that ever existed over the sample period for the regression analysis. We then subgroup mutual funds and unit trusts by types, i.e. equity, balanced, and fixed-income, and similarly assess their selection ability and market timing skills. We specify the conditional Jensen alpha, which is a direct derivation from the Capital Asset Pricing Model as:

(1)Rpt −Rft =αp +βp(Rmt −Rft) +εpt

Where R_pt is the return of portfolio p at time t; Β_p is the CAPM beta; R_ft is the risk-free rate of return at time t; α_p is the intercept or alpha; R_mt is the market return of the portfolio at time t; and Rmt − Rft is excess market return; εpt is the random return component of the portfolio. Our variable of interest αp indicates selection ability. A positive αp shows fund selection ability, while a negative αp portrays otherwise.

Treynor and Mazuy (1966) posit that if market timing exists, it induces curvature in the security market line (SML). The curvature in SML arises from the fund manager's increased investment in higher beta stocks when expecting a bull market. Similarly, the fund manager's anticipation of a bear market will reduce investment in higher beta portfolios by moving into ‘safer' securities. Treynor and Mazuy argue that adding a quadratic term in the CAPM shows the curvature induced in the SML. We thus specify the unconditional Jensen Alpha as:

(2)Rpt −Rft =αp +βp(Rmt −Rft) +γpt(Rmt −Rft)2 +εpt

Where γ_pt is the coefficient of the market timing skill of portfolio p at time t, all other variables are explained in equation (1). A positive γ_pt depicts a fund manager's market timing skill, while a negative γ_pt shows otherwise.

From equation (2), Treynor and Mazuy (1966) assume that risk premium is solely explained by variation in market return. Nevertheless, it is well established in the literature that multifactor models best explain security returns (Carhart, 1997; Fama & French, 1993, 2016, 2017). We employ unconditional versions of Carhart (1997), Fama and French (1992) multifactor models below to examine mutual funds and unit trusts' selection ability and market timing skills.

(3)Rpt – Rft =αp +βop (Rmt – Rft) +β1p SMBt +β2p HMLt +γpt (Rmt – Rft)2 +εpt

(4)Rpt – Rft =αp +βop (Rmt – Rft) +β1p SMBt +β2p HMLt +β3p UMDt +γpt (Rmt – Rft)2 +εpt

Where β_0,1,2,3 is the factor coefficient, SMB_t is the excess return between small-cap portfolio and large-cap portfolio (Small minus Big), HML_t is the excess return between value stocks and growth stocks (High minus Low), and UMD_t is the difference between the return of a portfolio of previous winners and a portfolio of prior losers (Up minus Down), i.e. momentum factor. All other variables are explained in equations (1 and 2). Our objects of interest in the above equations are the α_p which measures selection ability, and γ_pt, which indicates market timing skills.

We estimate the above equations using the OLS approach. Due to its distribution-free characteristics, the generalized method of moment (GMM) has recently gained popularity among researchers. Nonetheless, for linear models, the imposed moment criteria by OLS and GMM estimates are identical (Ferson & Schadt, 1996). We test for autocorrelation and heteroskedasticity in the model. The Durbin-Watson test for autocorrelation is 0.6699, which indicates positive autocorrelation, while the probability Chi-square for the Breusch-Pagan test for heteroskedasticity, is 0.6036, which fails to reject the heteroskedasticity null hypothesis of constant variance. We, therefore, deploy HAC-robust standard errors to the OLS estimations. HAC-consistent robust standard errors produce estimates equivalent to GMM (Agarwal & Pradhan, 2018; Fadikpe et al., 2022) and also help to address autocorrelation and heteroskedasticity issues that may be present in the model.

4. Empirical results and discussion

4.1 Descriptive statistics

Table 3 below shows that the study used 32 Mutual Funds (MF) and 17 Unit Trusts (UT). MF and UT's minimum and maximum values are −0.1256 and 0.8920, and −0.0850 and 0.7043, respectively. The mean values of MF_equity and GSE-CI are 0.1552 and 0.1311, with corresponding standard deviations of 0.2292 and 0.3274, respectively. MF_equity and GSE-CI have variations larger than their respective means. This indicates larger dispersion of returns for MF_equity and GSE-CI from their average compared to the rest of the funds. It thus suggests that investors should expect wide swings in returns when investing in MF_equity and the stock market. The wide deviations in GSE-CI and MF_equity returns are expected since investments in stocks and or stock-related funds have such attributes. UT generally has higher minimum and lower maximum values than MF, yet UT's mean return (0.1945) is higher than MF (0.1844). This shows that, on average, UT provides higher returns to investors than MF. Although deviations from the returns of MF (0.1581) and UT (0.1149) are not wider than their respective mean returns, MF deviates from its average return more than UT. Hence, investors seeking marginal dispersions from funds' average returns would be better off investing in UT. Tbills has the least standard deviation among the variables, suggesting that its return is the least volatile.

4.2 Results and discussion

The output of mutual fund (MF) and unit trust (UT) selection ability and market timing skills depicted in Table 4 shows no selection ability but evidence of market timing skills in each factor model deployed. Under the Carhart four-factor unconditional model, both MF (−0.0628) and UT (−0.0699) exhibit significant non-selection abilities at the 5% significance level. However, UT shows the highest significant non-selection ability. In terms of magnitude, the lowest non-selection ability occurs for MF in Panel A (under the Jensen alpha) and UT in Panel B (under the Fama-French [FF]). On market timing skills, all the unconditional models exhibit positive market timing skills indicating that both MF and UT managers can time the market. The highest market timing skills are in the unconditional FF three-factor model for MF (0.3770) and Panel A Jensen alpha for UT (0.3263), significant at the 10 and 5% significance levels, respectively. The highest R-squared for MF and UT happens in the unconditional Carhart four-factor model.

Across all the models, MF and UT have no selection ability, but they do possess market timing skills. Hence, mutual fund and unit trust managers cannot select the best-performing portfolios at a given risk level. Yet, they can accurately forecast the market's direction to adjust their portfolio exposures. MF has less selection inability but higher market timing skills across most models than UT. This result indicates that although fund investors may not enjoy selection ability in the Ghanaian collective investment scheme industry, they are better off subscribing to mutual funds since their selection inability is lower than unit trusts. Similarly, mutual fund managers exhibit higher market timing skills across most models than unit trust managers. The selection inability but superior market timing skills of fund managers in Ghana's collective investment scheme industry is akin to the findings of Thobejane et al. (2017), who found no selection ability but evidence of market timing skills in South Africa, but contrary to Agarwal and Pradhan (2018) and Tan (2015) who found selection ability and market timing skills, and no selection ability and no market timing skills respectively.

Table 5 displays MF and UT types' selection abilities and market timing skills. The unconditional Jensen alpha regression results in Panel A indicate that, apart from MF_fixedincome (0.0299) and MF_balanced (0.0043), managers of MF_equity and all unit trusts do not have selection ability. MF_equity (−0.0393) managers exhibit the highest significant selection inability at the 10% significance level. The types of mutual funds and unit trusts all exhibit market timing skills. But the market timing skill of UT_fixedincome is significant at the 10% significance level. By comparison, MF_equity, UT_balanced, UT_equity, and UT_fixedincome have the worst selection inability in that order. The above result indicates that investors would be better off investing in MF_fixedincome and MF_balanced than the other types of mutual funds and unit trusts since they can enjoy both the manager's selection ability and market timing skills.

Table 5 Panel B shows the result of the unconditional FF three-factor model. MF_equity (−0.0680, significant at the 10% significance level) and UT_balanced (−0.0111) have selection inabilities, while all the other types of mutual funds and unit trusts have selection abilities. MF_fixedincome has a superior selection ability compared to UT_fixedincome, while MF_equity has the highest selection inability in the industry. Again, all the types of mutual funds and unit trusts have market timing skills, with that of MF_equity being significant at the 10% significance level. Managers of MF_equity (0.5474) have superior and significant market timing skills in the industry, while MF_balanced (0.2869) has the lowest market timing skills. UT_equity has positive selection ability and market timing skills compared to MF_equity, which has the worse selection inability but significant market timing skills. While MF_balanced exhibits selection ability and market timing skills, UT_balanced has no selection ability but market timing skills. Both MF_fixedincome and UT_fixedincome have selection ability and market timing skills, but MF_fixedincome has superior selection ability, while UT_fixedincome possesses the highest market timing skills.

The unconditional Carhart 4-factor regression result is in Table 5 Panel C. All types of mutual funds and unit trusts do not have selection abilities. Selection inabilities of MF_balanced, MF_equity, and UT_balanced are significant at the 5, 10, and 5% significance levels, respectively. Apart from UT_balanced (−0.0277), which has no market timing skill, all the other types of mutual funds and unit trusts have superior market timing skills. The market timing skill of MF_equity is the highest in the industry, while that of MF_balanced is the least. While MF_equity has the worse selection inability but the highest superior market timing skill compared to UT_equity, MF_balanced has the least selection inability but superior market timing skill compared to UT_balanced. MF_fixedincome has the least selection inability and market timing skill compared to UT_fixedincome.

From the above results, MF_balanced and MF_fixedincome are the only funds that exhibit selection ability and market timing skills, as demonstrated by Panels A and B in Table 5. The co-existence of selection ability and market timing skills for MF_balanced and MF_fixedincome relate to the findings of Dong and Doukas (2019) for the European market. This result is also in tandem with Agarwal and Pradhan (2018), who demonstrate selection ability and market timing skills in the fund industry in India. The selection inabilities but superior market timing skills of MF_equity, UT_equity, UT_balanced, and UT_fixedincome are at variance with Das and Uma Rao (2015), who find selection ability but no market timing skills in the and Nikolaos et al. (2020) and Musah et al. (2014) who respectively document selection inability and no market timing skills in the Greek and Ghanaian fund industries. We may attribute the market timing skills to the managers' ability to envisage future movements of macroeconomic variables and the seemly weak-form efficiency of the Ghanaian market, which creates opportunities for market timing. The non-existence of selection ability exhibited by mutual funds and unit trusts managers in the Ghanaian fund industry may result from fund managers' lack of superior research expertise on specific stocks to form a portfolio or their inability to fully utilize the momentum factor in portfolio construction.

5. Conclusion

The study analyses the selection ability and market timing skills of 32 mutual fund and 17 unit trust managers in the Ghanaian fund industry from 2011 to 2019. We construct factor loadings for the market, enabling us to utilize multifactor models in our assessment. We deploy the unconditional Jensen alpha, the multifactor Fama-French (FF) three-factor, and Carhart four-factor models in our estimation. Based on the results from all the models engaged, we conclude that mutual funds and unit trusts exhibit no selection ability but have superior market timing skills. Managers can forecast macroeconomic variables in their portfolio formations but lack the skill to select portfolios to generate superior risk-adjusted returns. We also evidence that among the types of mutual funds and unit trusts, only managers of fixed-income and balanced mutual funds exhibit selection ability and market timing skills in the industry, as shown by most of the models deployed. We further document that all the types of unit trusts and only equity mutual funds have market timing skills but no selection abilities.

The selection ability and market timing skills of mutual fund and unit trust managers provide helpful information to investors to aid their decision-making on the type of fund to subscribe to. Also, fund managers' fund selection and market timing skills provide asset management companies the opportunity to evaluate the skills of their fund managers to proffer the needed training and upgrade in research tools for their respective managers. The study suggests that asset management companies boost their fund managers' research capabilities, which will help them improve their selection abilities. The study finally recommends that stock market regulators help increase the listing of companies on the market since the small number of listed firms on the Ghana stock market may be a reason for the lack of selection abilities of mutual fund and unit trust managers.

The study is limited to a single developing African economy which may hinder the generalization of the results. We also do not indicate whether market timing skills and selection abilities impact mutual fund and unit trust cashflows. These provide avenues for future research.

Table 1

List of mutual funds and unit trusts

Fund name	Category	Type	Name of manager
All-time Bond Fund	MF	Fixed-income	All Time Capital Ltd
Anidaso Mutual Fund	MF	Equity	New Generation Investments Services Ltd
CM Fund	MF	Equity	SCD Brokerage Ltd
Capital Growth Fund	UT	Balanced	IC Securities (Gh) Ltd
CDH Balanced Fund	MF	Balanced	CDH Ltd
Christian Community Mutual Fund	MF	Balanced	Black Star Advisors
Crystal Entrepreneur Fund Ltd	MF	Balanced	Crystal Capital Ltd
Crystal Wealth Fund	MF	Fixed-income	Crystal Capital Ltd
Dalex Vision Fund	MF	Balanced	Octane DC Ltd
Databank ARKFUND	MF	Balanced	Databank Asset Mgt Service Ltd
Databank Balanced Fund	MF	Balanced	Databank Asset Mgt Service Ltd
Databank Educational Fund	MF	Balanced	Databank Asset Mgt Service Ltd
Databank Epack Inv. Fund	MF	Equity	Databank Asset Mgt Service Ltd
Databank MFund	MF	Fixed-income	Databank Asset Mgt Service Ltd
EDC Ghana Balanced Fund	MF	Balanced	EDC Investment Ltd
EDC Ghana Fixed Income Fund	UT	Fixed-income	EDC Investment Ltd
EDC Kiddi Fund Mutual Fund	MF	Fixed-income	EDC Investment Ltd
EDC Money Market Unit Trust	UT	Fixed-income	EDC Investment Ltd
EM Balanced Unit Trust	UT	Balanced	EM Capital Partners Ltd
First Fund	MF	Fixed-income	FirstBanc Financial Services Ltd
FirstBanc Heritage Fund Limited	MF	Equity	FirstBanc Financial Services Ltd
Freedom Funds Unit Trust	UT	Fixed-income	Liberty Capital Gh. Ltd
Galaxy Balanced Fund Ltd	MF	Balanced	Octane DC Ltd
Galaxy Money Market Fund Ltd	MF	Fixed-income	Octane DC Ltd
Gold Fund Unit Trust	UT	Equity	Gold Coast Fund Mgt. Ltd
Gold Money Market Fund	MF	Fixed-income	Gold Coast Fund Mgt. Ltd
Horizon Fund	MF	Equity	NTHC Ltd
Legacy Unit Trust	UT	Fixed-income	IFC Capital Management Ltd
MC Ottley Unit Trust	UT	Balanced	McOttley Capital Limited
MyWealth Unit Trust	UT	Balanced	IFC Capital Management Ltd
Nordea Income Growth Fund	MF	Fixed-income	EcoCapital Inv. Mgt. Ltd
Omega Equity Fund	MF	Equity	Omega Capital Ltd
Omega Income Fund	MF	Fixed-income	Omega Capital ltd
Republic Equity Trust	UT	Equity	Republic Investments Ghana Ltd
Republic Future Plan Trust	UT	Balanced	Republic Investments Ghana Ltd
Republic REIT	UT	Equity	Republic Investments Ghana Ltd
Republic Unit Trust	UT	Fixed-income	Republic Investments Ghana Ltd
Richie Rich Unit Trust	UT	Fixed-income	IFC Capital Management Ltd
SAS Fortune Funds	MF	Equity	Strategic African Securities Ltd
SEM All-Africa Equity Fund	MF	Equity	SEM Capital Advisors Ltd
SEM Income Fund	MF	Fixed-income	SEM Capital Advisors Ltd
SEM Money Plus Fund	MF	Fixed-income	SEM Capital Advisors Ltd
Sirus Opportunity Fund	MF	Balanced	Sirus Capital Ltd
Stanlib Cash Trust	UT	Fixed-income	Stanbic Investment Mgt. Service Ltd
Stanlib Income Fund Trust	UT	Fixed-income	Stanbic Investment Mgt. Service Ltd
TTL Income Haven Fund	MF	Fixed-income	TTL Capital Ltd
UMB Balanced Fund	MF	Balanced	UMB Investment Holdings Ltd
uniSecurities Unit Trust	UT	Fixed-income	uniSecurities (Ghana) Limited
Western Oil and Gas Fund	MF	Equity	Weston Capital Ltd

Note(s): MF is mutual funds, while UT is unit trusts

Source(s): Authors’ compilation from Ghana SEC annual reports

Table 2

Annual factor loadings for Ghana from 2011 – 2019

Year	Rmt − Rft	SMB	HML	UMD
2011	−0.1340	0.0087	0.0352	0.3156
2012	0.0091	−0.1292	0.1315	1.1625
2013	0.6001	−0.3126	0.0722	3.3182
2014	−0.2039	−0.3085	0.2374	1.0592
2015	−0.3489	0.3603	0.2201	3.7450
2016	−0.3214	−0.0543	−0.1973	0.9476
2017	0.3940	−0.3730	−0.3378	2.1740
2018	−0.1485	−0.1535	0.2671	3.0814
2019	−0.2694	0.0453	−0.0884	1.1133
Average	−0.0470	−0.1019	0.0378	1.8796

Note(s): Annual 4-factor loadings [excess market return, Small minus Big (SMB), High minus Low (HML), and Momentum (UMD)] constructed with end-of-year trading data from the Ghana stock exchange from 2011 – 2019. Rmt − Rft is excess market return, i.e. return of the market [proxied by the Ghana Stock Exchange Composite Index (GSE-CI)] above the risk-free rate. Positive (negative) excess market return indicates that the stock market outperforms (underperforms) the risk-free rate. SMB is the excess return of small-cap portfolio and large-cap portfolio (Small minus Big). Positive (negative) SMB indicate that small (big) capitalization stock portfolios outperform big (small) capitalization stock portfolios in the market. HML is the excess return between high book-to-market stocks, i.e. value stocks, and low book-to-market stocks, i.e. growth stocks (High minus Low). Positive (negative) HML indicates that value (growth) stocks outperform growth (value) stocks in the market. UMD is the momentum factor which is the difference between the return of positive advancing portfolios and negative advancing portfolios (Up minus Down). Positive (negative) UMD indicates that returns of high (lower) advancing stocks influence returns in the market

Source: Authors’ computation

Table 3

Descriptive statistics

Variables	No. of funds	Obs	Mean	Std dev	Min	Max
MF	32	205	0.1844	0.1581	−0.1256	0.8920
MF_equity	9	67	0.1552	0.2292	−0.1256	0.8920
MF_balanced	11	69	0.1848	0.1170	0.0074	0.5276
MF_fixed-income	12	69	0.2123	0.0928	−0.0011	0.4700
UT	17	114	0.1945	0.1149	−0.0850	0.7043
UT_equity	3	27	0.1813	0.1727	−0.0266	0.7043
UT_balanced	5	26	0.1798	0.1093	−0.0850	0.3948
UT_fixed-income	9	61	0.2067	0.0815	0.0140	0.5443
Tbills	−	9	0.1781	0.0520	0.1030	0.2579
GSE-CI	−	9	0.1311	0.3274	−0.1533	0.7881

Note(s): The above table provides descriptive statistics of variables used in the study from 2011 to 2019. Obs is the number of observations. MF is mutual funds, while UT is unit trusts. MF_equity are mutual funds that are predominantly equity-focused, MF_balanced are mutual funds that focus on equities and fixed-income instruments, and MF_fixed-income are mutual funds that focus on only fixed-income instruments. UT_equity are unit trusts which are predominantly equity-focused, UT_balanced are unit trusts which focus on equities and fixed-income instruments, and UT_fixed-income are unit trusts which focus on only fixed-income instruments. Tbills denote treasury bills, while GSE-CI is Ghana stock exchange composite index which measures stock market performance

Source(s): Author’s computation

Table 4

Selection ability and market timing skills of Mutual Funds and Unit Trusts

2011 – 2019	MF	UT
Panel A: Jensen
Selection ability	−0.0079	−0.0171
	(0.01528)	(0.0171)
Market timing skill	0.2896*	0.3263**
	(0.1696)	(0.1485)
Obs	205	114
R-squared	0.3909	0.2699
Panel B: FF
Selection ability	−0.0289	−0.0026
	(0.0220)	(0.0232)
Market timing skill	0.3770*	0.2715
	(0.1921)	(0.1659)
Obs	205	114
R-squared	0.4091	0.3319
Panel C: Carhart
Selection ability	−0.0628**	−0.0699**
	(0.0248)	(0.0281)
Market timing skill	0.3192	0.1339
	(0.2108)	(0.1797)
Obs	205	114
R-squared	0.4151	0.7309

Note(s): MF is mutual funds, and UT is unit trusts. Obs is the number of observations. Standard errors are in curve brackets, while *, **, and *** depict 10%, 5%, and 1% significance levels, respectively

Source(s): Authors’ computation

Table 5

Selection ability and market timing skills of Mutual Funds and Unit Trusts types

2011 – 2019	MF_equity	MF_balanced	MF_fixedincome	UT_equity	UT_balanced	UT_fixedincome
Panel A: Jensen
Selection ability	−0.0393*	0.0043	0.0299	−0.0218	−0.0226	−0.0078 (0.0170)
	(0.0229)	(0.0253)	(0.0237)	(0.0377)	(0.0435)
Market timing skill	0.4207	0.1798	0.1208	0.4031	0.2206	0.3018*
	(0.2878)	(0.2533)	(0.1653)	(0.3273)	(0.2866)	(0.1607)
Obs	67	69	69	27	26	61
R-squared	0.6655	0.3781	0.1852	0.6217	0.2484	0.1616
Panel B: FF
Selection ability	−0.0680**	0.0001	0.0103	0.0021	−0.0111	0.0019
	(0.0331)	(0.0312)	(0.0339)	(0.0512)	(0.0510)	(0.0258)
Market timing skill	0.5474*	0.1826	0.1986	0.3082	0.2108	0.2620
	(0.3267)	(0.2869)	(0.1912)	(0.3765)	(0.2880)	(0.1840)
Obs	67	69	69	27	26	61
R-squared	0.6731	0.4164	0.2729	0.6369	0.5823	0.2080
Panel C: Carhart
Selection ability	−0.0587*	−0.0649**	−0.0040	−0.0142	−0.1451**	−0.0666
	(0.0317)	(0.0275)	(0.0582)	(0.0468)	(0.0536)	(0.0410)
Market timing skill	0.5765	0.1013	0.1510	0.2423	−0.0277	0.1707
	(0.3986)	(0.3064)	(0.1804)	(0.4947)	(0.2344)	(0.1786)
Obs	67	69	69	27	26	61
R-squared	0.6735	0.4499	0.3013	0.6394	0.7399	0.2846

Note(s): MF_equity is a mutual fund composed of equity instruments, MF_balanced is a mutual fund composed of equities and fixed-income instruments, and MF_fixedincome is a mutual fund composed of fixed-income instruments. UT_equity is a unit trust composed of equity instruments, UT_balanced is a unit trust composed of equities and fixed-income instruments, and UT_fixedincome is a unit trust composed of fixed-income instruments. Obs is the number of observations. Standard errors are in curve brackets, while *, **, and *** depict 10%, 5%, and 1% significance levels, respectively

Source(s): Authors’ computation

Note

1. Source: http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/Data_Library/f-f_factors.html The authors thank Kenneth R French for generously providing elaborate guidance on his website.

Funding: The authors did not receive any financial support.

Ethical statement: This article does not contain any studies with human participants or animals performed by the authors.

Availability of data and materials: The datasets on mutual funds and unit trust can be sourced from the Annual reports Ghana: www.annualreportsghana.com, and the official database of the Ghana SEC: https://sec.gov.gh/sec-annual-reports/. The stock market dataset for constructing the factor loadings can be purchased from Databank Research: www.info@databankgroup.com.

Conflict of interest: All authors state that there is no financial or non-financial interest to disclose.