Due to globalization, internationalization of large financial groups, and the good performance of stocks and long-term bonds, the investment fund markets grew exponentially in the 1990s (Klapper, Sulla, & Vittas, 2004). In Brazil, this expansion was also influenced by more specific reasons, such as inflation control after the Plano Real, an economic plan that stabilized the economy and the Brazilian capital market (Tizziani, Klotzle, Jr & Motta, 2010).

The Brazilian fund industry is considered the best positioned among the other emerging countries (Klapper, Sulla & Vittas, 2004; Varga & Wengert, 2011). The industry has improved, showing constant evolution in regulation, an increase in the number of qualified managers, and the variety of financial instruments available in the capital market. Thus, Brazilian investors have shown greater interest in the fund industry compared to traditional investments such as savings (Varga & Wengert, 2011).

ANBIMA’s reports (2019) suggest the funds’ industry growth in Brazil. The multimarket investment funds stand out, representing 21% of the total equity (BRL 1 trillion) of the investment fund industry in Brazil in May 2019. This growth led to changes in funds’ classification in 2014 in order to simplify the industry, facilitate understanding, and expand it to the general public. According to the Brazilian Securities and Exchange Commission’s instruction CVM 555 in December 2014, the funds were classified into four classes (level 1): fixed income, multimarket (balanced/mixed), stocks, and foreign exchange.

The investment funds are classified, in a second level, into three categories: allocation, strategies, and investments abroad. Finally, the classification advances to a third level (subcategories). Funds classified as strategies in the second level are separated into the following eight subcategories in the third level (according to strategies they adopt): protected capital, specific strategy, interest and currencies, free, long and short neutral, long and short directional, macro, and trading. The multimarket investment funds in the subcategory free strategy form the most comprehensive group of this class of funds, concentrating more than 7% of the entire net equity of the fund industry (ANBIMA, 2019).

Matos, Penna, and Silva (2015) found that, in Brazilian equity funds, the superior results are associated with lower management fees and collection of performance fees. The authors also observed an inverse relationship between the funds’ maturity and profitability. In multimarket investment funds, Maestri and Malaquias (2018) found evidence of better returns associated with new managers, with the possible explanation that younger managers suffer from the behavioral bias of overconfidence and have a lower level of risk aversion (Chevalier & Ellison, 1999; Li, Zhang, & Zhao, 2011).

In addition, Maestri and Malaquias (2018) suggest that better performance is associated with funds with a higher share of fixed income and funds with higher net worth. This evidence converges with the findings of William Rochman (2009), who observed better performance in stock funds and multimarket investment funds, with higher net assets and longer duration. The performance rate explains part of the superior performance, but not the increase in total risk (Ackermann, Mcenally, & Ravenscraft, 1999). As for performance, Filho and Sousa (2015) found that management fees were more relevant than performance fees and Treynor, Sharpe, and Modigliani indices.

Fonseca (2012) is one of the few works that cover the ANBIMA subcategories in Brazil. The author showed that the managers of long and short funds and balanced funds demonstrated market timing and selectivity abilities in the period from 2005 to 2010. Also, Fonseca (2012) observed better returns associated with multi-strategy funds and macro funds. This work is shown in the context of the new CVM instruction 555 and differs by measuring performance from the manager’s perspective of discretion and identifying possible conflicts of interest.

Managers may try to benefit from the asymmetry of information existing in the investment fund industry (Moreira, Tavares & Malaquias, 2017). Performance measures can be manipulated (Goetzmann, Ingersoll, Spiegel, & Welch, 2007), and there are several studies, such as Roquete, Maranho, Klötzle, and Pinto (2016), that try to understand the abnormal movement of resources prior to the half-yearly closing, which can be interpreted as the manager’s attempt to inflate stocks, even if artificially, in the period before the measurement of the performance rate.

Roquete, Maranho, Klötzle, and Pinto (2016) argue that there are agency problems in Brazilian multimarket funds. The authors found evidence of abnormal post-semester negative returns, suggesting a forced sale of positions taken prior to the half-yearly closing. The excessive turnover of portfolios can cause losses to the market due to the artificial inflation of asset prices, and brokerage costs are unnecessarily absorbed by investors (Wilner, 2000). Huang, Sialm, and Zhang (2009) report that stable funds perform better than funds with greater volatility and conclude that managers tend to act according to the incentive received.

In Brazil, Sanematsu (2013) found that attempts to inflate stocks at the end of the semester are common in funds directed to the general public, while in funds directed to qualified investors, managers tend to operate regularly in half-yearly closings, which shows the existing agency conflict. Chevalier and Ellison (1999) highlight that managers receive a fixed percentage of the fund’s equity in addition to performance fees. Hence, it is possible that at some point, a new non-explicit contract will be generated based on the flow-performance relationship. When considering the poor enforcement of this contract, the alignment of interests between investor and manager may be compromised.

Getmansky (2012) studied the life cycle of hedge funds, finding a positive relationship between performance and fund size. The author also observed that the fund’s survival is tied to being in the right category at the right time. Therefore, managers will likely capture the best opportunities according to market movements. The free strategy fund has this premise, so it is necessary to understand whether managers can extract wealth at different times in the economy.

Agarwal, Daniel, and Naik (2009) studied the role of managerial incentives and discretion related to hedge funds’ performance. The authors associated the manager’s discretion with the fund’s lockup period (grace period) and the redemption period (or the frequency with which investors can withdraw from the fund - either at the end of a determined period or with prior redemption notice, which is the time required to liquidate the shares and for the investor to withdraw the money). Agarwal, Daniel, and Naik (2009) found a positive relationship between managerial discretion and the performance of North American Hedge Funds.

Based on these results, managers with greater discretion to make the fund’s strategic decisions should achieve better performance. Thus, we present the following hypothesis:

H₁: Free strategy funds perform better than funds with a defined strategy.

Funchal, Lourenço, and Motoki (2016) found evidence that qualified investors are associated with lower risks while managers of funds aimed at the general public opt for greater risk exposure. There may also be an agency conflict between managers and shareholders regarding risk-taking. The authors argue that closed-end funds, which provide greater discretion to the manager, had a higher level of risk, which may be related to the fund’s long-term strategy.

Given that managers with greater discretion must produce better performance, it is crucial to understand whether this freedom results in less risk to the fund since managers are not tied to fixed strategies. Thus, we present the following hypothesis:

H₂: Free strategy funds present a lower level of risk than funds with a defined strategy.

The study was developed using a pooled OLS multiple regression technique with monthly fixed effects.

The econometric model used for the study was:

Where Performance^j_(ⅈ,t) = {gross return,Sharpe,Sortino} and Risk^j_(ⅈ,t) = {Volatility,DR,VaR,CVaR} are the risk measures for each category of fund ⅈ in month t. The variables of interest are represented by VarInt t^k_(ⅈ,t)They correspond to the subcategories free, protected capital, interest and currencies, long and short neutral, long and short directional, macro, and trading. The control variables, represented by Control t^l_(ⅈ,t) sare presented in subsection 3.3. Finally, the regression error is added.

The free strategy multimarket fund was used as a model reference, so its effect was absorbed by the constant. When using the free strategy subcategory as a basis, all the regression results for the other subcategories of funds are expressed in relation to the free strategy fund. In addition, this subcategory provides greater discretion to the manager and, in theory, brings the best benefit to investors (better performance and lower level of risk).

Despite the variety of benchmarks that multimarket funds may have, the interbank deposit rate (CDI) was used to calculate performance and risk measures such as the minimum return required by the investor and the risk-free rate of return. According to Varga (2001), it is possible to use the CDI as a risk-free rate in the market because it is the most appropriate asset in Brazil and has adherence to the profitability of federal public securities linked to the country’s reference tax rate (Selic).

Dependent Variables for performance: gross return (which measures the return discounting the management fee), Sharpe (measures the risk and return ratio based on the standard deviation), and Sortino (complements Sharp by measuring the risk ratio and return based on downside risk). Risk variables: downside risk (which measures the negative volatility of the assets), VaR (measures the maximum loss in normal market situations), CVaR (complements VaR, measuring the risk beyond the confidence interval), and Volatility (calculated by estimating the standard deviation of the fund’s historical returns). Details on the methodology for calculating the dependent variables are available in Appendix A.

The existing literature guided the inclusion of a set of fund characteristics as control variables that may affect the fund’s performance. Figure 2 (Appendix B) shows the aspects that can influence the fund’s performance and risk.

Several control variables were adopted, such as management fee, performance fee, shareholders’ equity, grace period, and closed-end funds. The methodology for calculating the variables is available in Appendix B.

Table 1 shows the distribution of multimarket investment funds according to the ANBIMA classification and their representativeness in absolute and percentage values. The free strategy funds are predominant, representing approximately 70% of the frequency in the distribution of the entire category of multimarket strategies.

Table 1
Distribution of Multimarket Investment Funds - ANBIMA Classification/subcategories

Source: elaborated by the authors based on data from funds of the Quantum Axis database.

Table 2 shows the descriptive statistics of the performance and risk variables that were 0.50% winsorized in each tail.

Table 2
Descriptive Statistics of Performance and Risk Variables

Source: elaborated by the authors based on fund data from the database Quantum Axis. Note: The data presented were winsorized to 0.5% for each tail. The interbank deposit (CDI) rate was adopted as the risk-free rate..

The results obtained and presented in Table 3 answer the hypothesis that managers with greater operational freedom obtain better results since the free strategy funds demonstrated better performance and lower risk than the other funds with fixed strategies.

Table 3
Estimates of performance and risk determinants

Source: Elaborated by the authors based on fund data from the database Quantum Axis. Notes: The table presents estimates for the OLS model pooled OLS regression with monthly fixed effects and the effect of the funds classified in the free strategy subcategory was absorbed in the model constant. Standard errors clustered by fund (N = 100,995), 3,499 different funds measured in the Brazilian market between 2016 and 2019. The dependent variables are gross return, Sharpe index, Sortino index, Volatility (Volat.), Downside Risk (DR), Value at Risk (VaR), Conditional Value at Risk (CVaR). The t statistics are in parentheses. *, **, and *** indicate statistical significance at 10%, 5%, and 1%, respectively. All dependent variables winsorized at 0.5% on each tail. The variables of interest are dummies, and correspond to the subcategories of multimarket investment funds: protected capital is 1, specific strategy is 1, interest and currencies is 1, long and short directional is 1, long and short neutral is 1, macro is 1, trading is 1. Management fee (%) is the fee charged by the fund’s manager per year. Performance fee is the fee paid as a percentage of the fund’s return that exceeds the variation of a previously determined performance index. Net redemption is the minimum term for the redemption of capital in days. Redemption rate is the rate levied on the amount redeemed when redemption occurs before the term established in regulation. Redemption fee (%) is the fee levied on the amount redeemed when redemption occurs before the term established in regulation. Nl (net equity) is the natural logarithm of the fund’s net worth in BRL. Age (months) is the age of the fund in months, on the date the data were obtained. Leverage is 1 if it is a leveraged fund. Big4 is 1 if fund is audited by Big 4 audit companies (Deloitte, Ernst and Young, KPMG, or Pricewaterhouse Coopers). Closed-end is 1 when it is a closed-end fund. Exclusive pension is 1 if it is a pension investment fund. Fund with high-water mark is 1 and 0 for the fund without high-water mark. Grace period is 1 if the fund has a grace period. Benchmark is 1 when the benchmark was used to calculate the fund’s performance. Exclusive to managers is 1 when the fund is destined to managers. Exclusive investor is 1 if the fund is destined to only one shareholder. Institutional investor is 1 if fund destined to institutional investors. Private Investors is 1 if a fund is intended for private investors. Professional investor is 1 if fund destined to professional investors. Qualified Investor is 1 when the fund is destined for qualified investors.

The specific strategy fund is the most static among the subcategories of funds, where the manager has a lower level of discretion since, at the time of launching, the objective and how to achieve it are defined. Therefore, it is possible to say these funds are the antithesis of the free strategy funds. When comparing these two subcategories, the specific strategy funds have statistically significant levels of risk. As for performance, it was significant only in Sharpe. Thus, it is possible to assume that when fund managers have more discretion, the fund presents a lower level of risk.

When comparing the free strategy fund with the protected capital fund, we see worse levels of performance and risk that were significant in Sortino and Conditional VaR in this category. As for risk, the result is contrary to what was expected since the fund has the premise of protecting part or all net worth. Thus, further studies are needed to explore better the protected capital funds’ promises regarding risk mitigation.

Contrary to what was found by Fonseca (2012), the subcategory macro did not result in high risk when compared to the free strategy. However, it presented worse performance when measured using Sortino. This type of fund makes medium and long-term investments, so the manager has less discretion than the free strategy funds. It is possible to assume that the result found by Fonseca (2012) was influenced by the macroeconomic environment of the period (2005 to 2010). As the data analyzed in this study comprised the period from 2016 to 2019, it is possible to assume that the low level of manager’s discretion significantly affected the fund’s performance.

The funds in the subcategory trading, in turn, were not significant in any of the indicators. There was an expectation that this category would be more adherent to the results of the free fund, as the manager is conducting short-term operations, taking advantage of market fluctuations. It is possible to consider that their level of discretion is higher than the others presented so far. In addition, it is characteristic of trading funds to use derivatives for leverage purposes, which translates into greater managerial discretion. In this situation, according to Nanda, Narayanan, and Warther (2000), high and significant levels of risk were expected, but no relationship was found.

The funds in the interest and currencies subcategory also underperformed the free strategy funds, as evidenced by gross return, but showed better risk indices in most indicators, as expected. The manager of this subcategory is limited to the interest and currency market and is exposed to fewer risk factors.

The long and short neutral funds showed a performance below the free strategy funds in gross return and Sortino, but demonstrating lower risk indices in all indicators. Risk is in line with expectations, as the manager of this subcategory considers similar assets in long and short positions to neutralize the risks. Finally, in the funds classified in the long and short directional subcategory, the manager weighs the parts bought and sold asymmetrically, presented a higher gross return. However, the Sortino index is lower in comparison with the free strategy. No relationship was found in the risk variables.

This data set may lead to the assumption that the manager’s discretion is important for the fund to perform well. The results corroborated the international literature, as observed in Agarwal, Daniel, and Naik (2009) when studying the behavior of North American Hedge Funds and concluding that funds, where managers have more discretion, generate better results. Regarding risk, it is not possible to state that free strategy fund managers can offer a lower level of risk than managers operating in funds classified in other subcategories. This happens because some multimarket investment funds are exposed to fewer risk factors than those observed in funds classified in the free strategy subcategory. This statement can be supported by the results found in the subcategories long and short neutral and interest and currencies.

Corresponding author: Tel. +55 (27) 99234-5441 E-mail: brunostorckk@gmail.com(B. A. Storck); fmotoki@gmail.com( . Y. S. Motoki) Fucape Business School. Av. Fernando Ferrari, 1358, Boa Vista - Vitória/ES - 29075-505, Brazil.