SSRN-id1872125

Electronic copy available at: http://ssrn.com/abstract=1872125

1

Active ETFs and their performance vis-à-vis passive ETFs, Mutual

Funds and Hedge Funds*

Panagiotis Schizas

Lecturer of Finance

Department of Economics

University of Peloponnese

22 100 Greece

Email: [email protected]

6/2/2011

Abstract

This work presents empirical results on the first active exchange traded funds (ETFs)

based on risk, return and incentives. Using models for both the returns and the volatility

of the underlying assets, I compare the performance of the suggested models with

alternative investment solutions such as passive ETFs, mutual funds and hedge funds.

The results show that active ETFs are not as active as they are considered by market

participants. The link between active and passive ETFs is strong, however the difference

in performance and risk weighs on the side of the passive ETFs. The results indicate also

that, in many cases, the active structure is surpassing mutual funds in terms of returns.

Finally, there is a unidirectional relation between active funds and hedge funds, since the

former is influenced by the latter.

Keywords: Active ETFs; Passive ETFs; Mutual Funds; Hedge Funds; VAR.

* The author appreciates helpful and insightful comments and suggestions by Gary Gastineau and

professor Dimitrios Thomakos.

Electronic copy available at: http://ssrn.com/abstract=1872125

2

Introduction

The decision in March 2008 by the Securities and Exchange Commission to

approve the listing of Active Exchange Traded Funds (active ETFs) in the US market

opened up a new form of asset management. Exchange-traded funds1 are considered a

passive investment solution that combines the dynamics of index-tracking unit trusts

with the merits and tradability of listed investment companies. Its structure surpasses the

major demerits of the aforementioned two vehicles given lower operating expenses,

trading liquidity, and more efficient tax structures than the conventional index-tracking

mutual funds. The launch of the active ETFs faces the challenge to maintain those

distinctive properties and let investors gain access to actively managed portfolios.

Nevertheless, the unique merit of full transparency of a passive ETF becomes the main

drawback of the new investment structure due to the obligation of mandatory disclosure

of the constituent holdings in addition to any re-balancing in real time.2

Full transparency means the daily disclosure of the entire portfolio and presents

the problem of “front running”, since the disclosure in real time of the portfolio

allocation could lead investors to replicate the allocation more promptly than the fund.

As a result, fund managers in the real world are reluctant to disclose their allocation in

real time. On the other hand, incomplete knowledge of the constituents of the

underlying portfolio violates the process of proper trading and could lead the market

maker to be misinformed and, consequently, fail to provide a “fair” price to the

investors3. Gastineau (2001) proposed the creation of a hedged portfolio as a proxy with

identical risk profile, so the market makers, specialists, investors and arbitrageurs can be

aware of risk exposure. The success of the new active structure depends on the ability to

overcome the aforementioned obstruction.

There is a lack of related empirical evidence on the relative performance of active

ETFs. Rompotis (2009), (2010) examined the performance and the bid-ask spread of the

first 4 active ETFs. On the contrary, there is extended literature based on passive ETFs,

1 Passive ETFs are considered a highly liquid passive worldwide investment strategy. There are

3,649 Exchange Traded Funds (ETFs) and Exchange Traded Products (ETPs) with 7,610 listings, assets of US$1,542.7 bn from 174 providers on 52 exchanges around the world (Blackrock ETF Landscape (March 2011)). 2

As the legislation calls for the investors and market-makers information, stock exchanges are obliged to

publish the indicative NAV every 5 seconds. In turn, dissemination of NAV requires the full knowledge of the underlying portfolios. Dissemination of NAV was the major obstacle, due to which 18 years passed after the inception of the first passive ETF before an actively managed ETF was born in the US. 3 There is a compulsory narrow spread between NAV and the floating price that specialists and market

makers should follow. The prompt disclosure benefits mainly professional investors since they are able to replicate the identical allocation, without paying ETFs expenses.

3

such as the ability to mimic the index (Gastineau (2004), Frino et al (2004), Elton et al

(2002)), the correlation between NAV and trading prices ((Cherry (2004), Delcoure and

Zhong (2007), Tse and Martinez (2007)) and the existence of tax efficiency characteristics

(Poterba et al (2004)). Pennathur (2002), Hughen et al (2007), Harper et al (2006)

investigated the different properties of ETFs and closed end funds, Tse et al (2006) and

Hausbrouck (2003) compared the spot and futures ETFs and Alexander et al (2007) and

Cabrera et al (2009) examined the predictability of the ETFs. Gleasona et al. (2004)

examined ETFs in periods of market instability and high volatility. Hendershott et al

(2005) examined ETFs and alternative trading venues, Boehmer et al (2003) and Hedge

et al (2004) investigated liquidity effects and Ascioglu et al (2006) compared ETFs and

common stocks.

This paper has three main objectives in its examination of active ETFs. The first is

to explore the structure of an active ETF and the obstacles that arise by the inception of

the new investment tool. The second objective is to examine and evaluate the

performance and the risk profile of the investment. Finally, to present its similarities and

the differences compared to alternative investment options such as passive ETFs,

conventional mutual funds and hedge funds.

Results show that current active funds have not as active as the market participants

expected, as the tracking error from the passive funds is low. The differences in

allocation between the active and the passive structure points out that active ETFs fail to

time the market as the latter appear to be less profitable and more volatile. The relative

comparison between active ETFs and hedge funds exhibits a unidirectional impact

deriving from the hedge funds. The analysis of active ETFs found they outperform

conventional mutual funds in the sense of mean returns.

The structure of this paper is as follows: Section 2 presents the ETFs

characteristics, section 3 gives a description of the database and the variables, and section

4 describes the methodology. Section 5 presents the main empirical results of the active

ETFs. This section also demonstrates a comparative analysis between the active ETFs,

the passive ETFs, the mutual funds and the hedge funds. Section 6 provides conclusions.

From Passive to Active ETFs: the similarities, the differences

and the obstacles

4

Active ETFs do not passively follow prescribed benchmark as passive ETFs do

and instead employ actual portfolio managers who either use rules-based strategies or

make portfolio calls based on a stated qualitative strategy. The legislation-imposed rule of

full transparency obliges that any amendment to the ETF should be disclosed promptly,

so that all participants are aware of the risk of the underlying portfolio.4 However, full

transparency in an active ETF creates several negative consequences, while, the major is

the “front running problem”. A solution that in practise clearly separates active and

passive ETFs could be named as “exemptive relief” and “blind trust”. Exemptive relief

means periodic portfolio disclosures identical to those in place for mutual funds, so every

three months with a specific lag. Blind trust working on behalf of an authorized

participant (AP) keeps the holdings of the portfolio hide. So, the blind trust becomes the

vehicle of the creation and redemption mechanism. The availability of real-time pricing

information will allow market participants to hedge trading exposures in shares

effectively and permit the efficient trading of shares in the market place without the need

for daily disclosure of the fund‟s portfolio holdings. Moreover, the blind trust could

hedge the funds for the exact cash value of the fund‟s value. So, creation units will be

created exclusively by the deposit of cash and will be redeemed by distributing securities

of the fund‟s portfolio to a blind trust that will liquidate securities in accordance with

instructions from the authorized participant redeeming shares. Furthermore, the

liquidation of ETF shares, will come up as cash to the AP, which never knowing what

made up the ETF shares that the blind trust redeems. Consequently, the adviser and the

fund subadviser are totally independent entities to an authorized participant to appoint in

transactions under the act. The proposed structured do not lose tax efficiency

characteristics of the passive ETFs since the in-kind process of creations and

redemptions takes place in the blind trust which is able to avoid imbedded capital gains.

Moreover, this structure eliminates the cash drag problem.

A different aspect is the predefined allocation of the passive ETFs, since investors

are no longer concerned with the dilemma which exchange traded fund, under the same

strategy, to prefer. On the contrary, investors on active ETFs base their decisions on the

manager‟s ability to generate profits. Therefore, the resignation of the fund manager

4 By legislation the stock exchanges are obliged to publish the INAV - indicative net asset value- every 5

seconds in order for the market maker to be informed about the ratio between the ETF and the underlying index. For the index linked ETFs, the legislation restricts an ETF to fluctuate more than 3% (including taxes) from the underlying index.

5

often has negative impact on the final performance of the fund as a result of a different

allocation.

Dataset Characteristics

There are limitations in this work since there are only five active ETFs in the

market with an adequate number of observations. The first four active ETFs launched

on April 2008 while the last one was added after its inception on November 24, 2008.

The data span extends from April 16, 2008 to March 4, 2010. More precisely, the funds

are:

● Active Low duration ETF (ticker: PLK), tracks Barclay‟s Capital 1-3 Year US treasury

index. The fund was launched on April 14, 2008 and has 7.62 million worth of assets

under management. The respective passive ETF that applied for the relative

comparison is Ishares Barclays 1-3 years (ticker: SHY).5

● Active Mega Cap ETF (ticker: PMA), tracks S&P500 index. The fund was launched

on April 14, 2008 and has 3.48 million worth of assets under management. The

respective passive ETF is SPDR S&P500 (ticker: SPY).

● Active AlphaQ ETF (ticker: PQY), tracks Nasdaq 100 index. The fund was launched

on April 14, 2008 and has 19.32 million worth of assets under management. The

respective passive ETF is Powershares QQQ (ticker: QQQQ)

● Active Alpha Multi cap ETF (ticker: PQZ), tracks S&P 500 index. The fund was

launched on April 14, 2008 and has 4.46 million worth of assets under management.

The respective passive ETF is SPDR S&P500 (ticker: SPY).

● Active US real estate ETF (ticker: PSR), tracks FTSE Nareit equity index. The fund

was launched on 21 November 2008 and has 11.53 billion worth of assets under

management. The respective passive ETF is Ishares FTSE Nareit real estate 50

(ticker: FTY). All the aforementioned active ETFs are provided by Powershares.

Clearly, the first active ETFs have not yet been matched by asset flows. Most

likely, the lack of interest is associated, with a lack of a well established track record,

which makes many advisors not to steer investors to them. As such, it's less likely that

active ETF assets will grow rapidly until customary track records are established.

Moreover, the first active ETFs could easily argue that follow traditional investment

strategies. The trading properties of the first active ETFS underline the imposed

5 Passive ETFs are the biggest funds based on market capitalization with respect to identical investment

strategy.

6

constraint of limited number of trades (not to exceed three trades per week). The

requirements of prompt disclosure that legislation imposes can be merely alleviated by

two specific solutions. The first solution refers to the option to hidden portfolio

reshuffling for the event day. In practise, the execution of the reshuffling could occur

every Friday. The manager has the time to mark and reveal his investment strategy before

the next business, on Monday. Unlike equity index ETFs, front running declines on

ETFs with fixed income strategies, which are not so easy to conquer to arbitrage

activities and may trade more frequently. Further to this point the next generation of

active ETFs are focused on the fixed-income, money markets, and global macro.

In order to compare the active ETFs with the mutual funds, I collected the mutual

funds‟ data in the following way: For each active ETF, I found the corresponding

Morningstar category6, I downloaded the fifteen biggest mutual funds of each peer group

and then calculated the daily mean return of each group. According to the Morningstar

categorisation for each active Fund PSR stands for Real Estate, PMA for large blend,

PLK for short government bonds, PQY for large growth and PQZ for mid-cap growth.

The horizon of the data span is identical to the active ETFs.

The data sample for the hedge funds is a result of search in the CISDM/hedge

funds database. In this work, I only used the hedge funds that include in their strategy

the key words „government bonds‟, „equity with exposure to the US market‟ and „real

estate‟, thus aligning with the categories of the active ETFs. The second step was to

calculate the mean return of each peer group. As it is known, hedge funds are organized

in monthly data and there is a delay in the available observations, so, while the data span

starts in April 2008, it ends in April 2009.

Methodology

This section presents and discusses the methodology and the approach that was

implemented in this paper. The first equation relates the return of the active ETF tar to

the age variable (AGE) which represents the number of the months since the inception

of each ETF7, the trading activity (TA) which stands for a dummy variable that takes the

value of one if there is a trading activity (it is otherwise zero) and the turnover variable

(TO) which stands for the daily log turnover of each fund.

6 In the prospectus of each ETF the Morningstar category to which it belongs is stated.

7 The same variable has been applied to examine hedge funds behaviour as they mature by Ackermann et

al. (1999).

7

ttttta TOaTAaAGEaar 3210 (1)

The next test tries to examine the twenty-day moving correlation tcory ,

between

the active and passive ETFs, with the inclusion of a lagged dependent variable, a dummy

variable as well as the cross-term.

, 0 0 1 , 1 1 , 1cor t t cor t cor t t ty a I a y y I (2)

,( )t t d bI I r c (3)

where ,( )t t d bI I r c

is a dummy variable which stands for an indicator for lagged

negative return of the benchmark - c being a fixed threshold, which in the estimations is

considered as zero. The estimations test for non-linearity and structural breaks. The

intuition behind this test is very crucial since the presence of any asymmetric response

points out that the correlation of the active structure strengthens more than the

correlation of the traditional passive ETFs.

In the second section, three different empirical equations are examined, where the

dependent variable is either the return of the active ETF tar or the return of the passive

ETFtpr . Equation (4) examines the impact of the Fama-French factors on the return of

the active ETFs, the passive ETFs and the mutual funds. The independent variables are

the market factor (M), which stands for the value weighted market excess return, the

(HML) variable, which stands for the book-to-market portfolio of high minus low

stocks, the (SMB), which stands for the size of the portfolio based on small equities

minus big equities and the (MOM), which stands for the portfolio of year long winners

minus year long losers.

tttttta MOMSMBaHMLaMaar 43210 (4)

The last part of this section examines through an OLS regression if standard risk

factors affect the returns and the risk of the active and passive ETFs. The risk factors

are:

DVG: The DRussell Value-Growth factor is expressed by the daily log difference

between the Russell 1000 value and the Russell 1000 growth index

HFI: the hedge fund index stands for the daily log difference of the generic hedge

fund index

DSL: The DRussell Large-Small factor is expressed by the daily log difference

between the Russell 3000 and the Russell 2000 index

8

VIX: the volatility momentum factor is expressed by the daily log difference of the

VIX index

VOL: the volume variable is a dummy variable which takes the value of 1 if the

trading volume is greater than the trading volume of the previous day, otherwise

zero.

ttttttta VOLaVIXaDSLaHFIaDVGaar 543210 (5)

In the third section, I consider the following unrestricted vector autoregressive

model for the return of the active and passive ETFs against the return of their common

benchmark. The order of both models is selected by the AIC criterion. So,

tp

ta

t r

ry (6)

tbt rx (7)

where they are formed into a system as follows:

t t tL y x u

(8)

where

p

a

, and similarly for

p

a

and

tp

ta

t u

uu (9)

Then, to check for any “alpha” asymmetries between active and passive ETFs I apply a

Wald test to check if the means of the two assets are identical both in the short run

means and in the long run means.

pao aaH :)1( (10)

paoH :)2( (11)

where, a

p

(12)

and 1

)1(

(13)

where (12) and (13) check for any dissimilarities into the long run means. Also, the

methodology, as presented by equations (6) to (13), is applied between the means of the

active ETFs and the respective mean of each group of the mutual Funds.

9

The last part of the third section, examines asymmetries between the return of the

active ETFs and the hedge funds, but in the context of panel data due to the lack of

sufficient observations of the hedge funds. The dependent variable is either the return of

the active ETF tar or the return of the hedge funds hftr , and the estimations

experimented with the inclusion of the lagged values of the active ETFs and the hedge

funds and the contemporaneous values either of the active ETFs or the hedge funds.

thfthftatta RaRaRaar ,13,2,110 (14)

tattahfthft RaRaRaar ,13,2,110, (15)

In equations (14) and (15) I applied a panel OLS cross section regression with

fixed effects and GLS weights and then I checked for any “alpha” asymmetries through a

Wald test as in (10).

Summary characteristics of the Active ETFs

Table 1, represents the basic features and descriptive statistics of the active ETFs.

[Insert Table 1]

Table 2 investigates, through an OLS regression (1), the relation between the

return and the risk of each ETF against a set of variables in order to check for

abnormalities related with the infant period of the funds8. The estimations indicate that

PMA, PQY and PQZ are positively linked to the age variable. This behaviour can be

explained as the investors have alternative solutions and increase their interest only as the

funds mature. The return of PMA was found to be negatively related with the turnover

variable.9

[Insert table 2]

Table 3 represents the OLS estimations of 5-day moving correlation of equations

(2) and (3). The results show that the current correlation is significantly related to the

lagged correlation across the ETFs.

[Insert Table 3]

8 I conduct the same OLS estimations using the daily volatility of the funds as independent variable, but

the variables do not have any influence on the volatility. 9 The low AUMs and the wide bid-ask spreads influence negatively the final performance of the ETFs and

consequently avert inventors to trade in. Moreover, professional investors are reluctant to invest since they only implement their strategy for small amounts.

10

Table 4 shows the basic descriptive statistics of the active ETFs segmented by

different time horizons. This segmentation helps to compare the outcome during

different periods and investigates the behaviour of the funds as they are growing up. The

mean and the median values of the active ETFs are in majority worse compared with the

respective values of the passive funds. The risk of the active ETFs is found to be less

volatile than the risk in the passive ETFs, however the diversification benefit is not large

enough to compensate investors, as can be extracted from the worse Sharpe ratio.

Among the equity active and passive ETFs, the active funds are found to be negative

skewed with excess kurtosis while the passive ETFs are positive skewed, exposing less

excess kurtosis. PMA is the least volatile fund among the active ETFs. The passive

structure outperforms the active structure, contrary to expectations. Focusing on the

bond sector, PLK is the laggard in all the basic indicators and becomes more noticeable

in the relative comparison of the Sharpe ratio.

Panel B examines the tracking error of the active versus the passive ETFs10

. The

results show that active and passive structures do not demonstrate any significant

diversification which is contrary to expectations. The most diversified active fund

appears to be the PSR which diverges 2.37 percent against the respective passive fund.

[Insert Table 4]

Active ETF structure and relative performance

This section performs a series of estimations to compare the performance of the

active ETFs versus the passive ETFs and the mutual funds. Table 5 provides the results

of equation (4) between the active and the passive ETFs, the mutual funds and the Fama

and French risk factors. Daily active return appears to be statistically significant with the

market factor, however, overall, less related than the return of the passive ETFs. PLK

and PSR exhibits the exceptions. Comparing the three equity funds -PMA, PQY and

PQZ- only PMA is negatively influenced by the size factor and the momentum trend of

the market. The results of the pair PSR-FTY show a strong positive relationship with the

size of the market. Mutual funds found to be uncorrelated both with the Fama and

10 The tracking error has been calculated as in Frino and Gallagher (2001) by estimating the average of absolute differences between the returns of active ETFs and the returns of the corresponding passive funds. This type of tracking error takes into account the absolute value of return differences which means that either a positive or a negative deviation points out a performance deviation.

11

French factors. The only exception comes out from the HML factor in PSR-FTY pair,

where there is a positive link to ETFs and negative link to mutual funds.

[Insert Table 5]

Figure 1, shows the 5-day moving correlation between the PQZ and QQQQ and

the movement of the Nasdaq 100 index. The results indicate two crucial values, 0.5 and

0.0, as can be easily extracted from the results where the correlation between the active

and passive funds collapses. The figure outlines that as the index moves to the outliers,

the correlation diminishes. Table 6 represents the pairwise correlation for three different

periods in order to check for any clustering as I extend the estimation period. The results

point out that the bond sector‟s pair exhibits the biggest independence, while in contrast

there is the pair PQZ-QQQQ. On average, the equity ETFs exhibit strong correlation

which ranges between 0.58 and 0.83.

[Insert Table 6 & Figure 1]

Table 7 provides daily evidence of the style and market capitalization analysis of

both the active and passive structure. Estimations report the run of an OLS regression of

equation (5). Panel A provides evidence of the daily log returns against only the factors

that were found to be significant, while estimations did not include the bond sector‟s

ETF (PLK). The vertical analysis shows that PQY and PQZ are the only active funds

that indicate the existence of alpha, however, the daily excess return is half than the

respective return of the passive funds. PMA exhibits zero excess return, unlike the

respective passive ETF. PMA and PSR align with the behaviour of the passive funds,

which are found to be positively related to the style factor. The pairs PMA-SPY, PQY-

QQQQ and PQZ-SPY are strongly related to the hedge fund index and the volume

factor. However, PQY and PQZ are affected in a different direction by the daily volume.

The pair PQZ-SPY is affected negatively by the size factor while controversial results

arise between PMA and PQY, contrary to the passive funds. The fit of the regression is

greater in the passive structure than in the active structure, which confirms the

hypothesis that the active structure is less affected by the traditional factors.

Panel B tries to capture the existence of any patterns of the daily volatility between

the active and passive structure. The estimations show that pair PMA-SPY is negatively

related with the hedge fund index and the pair PSR-FTY is negatively related with the

12

Drussell large-small variable. Comparing Panel A and Panel B, the funds are strongly

affected by the daily trading volume.

[Insert Table 7]

Active versus passive

Table 8 shows the results of an unrestricted VAR estimation of equation (6), with

endogenous variables the returns of the active and the passive ETFs, and as exogenous

variable the daily return of the respective benchmark11. I found that the benchmark‟s

magnitude is longer in the active funds than in the passive funds. I will now discuss each

pair briefly. The pair PQY-QQQQ exhibits mean reversion for the active fund but not

for the passive. There is also feedback between the active and passive funds, as indicated

by their lagged values; however, the feedback is positive for the active ETFs and negative

for the passive ones. The benchmark is found to be insignificant. The pair PQZ-SPY

exhibits mean reversion only for the active ETF. There is strong positive feedback from

SPY to the PQZ; however, the feedback lies on the positive direction with the feedback

of its own lagged value. There is also high, almost one-to-one, influence of the

benchmark on both active and passive ETFs. The pair PSR-FTY has mean reversion for

both the active and passive fund. There is no link between the active and passive ETFs,

but both are affected by the benchmark. As expected, this effect is greater in passive

ETFs than in active ETFs. The pair PMA-SPY exhibits mean reversion for both types of

funds. There is a strong positive feedback from SPY to the PMA, although the passive

fund is found to be independent from the active. In addition, there is a strong influence

from the benchmark on both active and passive funds; however, the influence is almost

double the one expected in the passive funds. The pair PLK-SHY is found to be

unrelated, although it does exhibit mean reversion on the active ETF. Both funds are

strongly linked to the benchmark, even though the passive fund is found to be more

correlated.

[Insert Table 8]

In table 9, I compare the short and long run mean of the active and the passive

funds as indicated in (8) and (9) through a Wald test. The results indicate that in both

11

According to the AIC information criterion, the efficient lag intervals are found to be one.

13

tests, excess returns are identically zero, with p-values that are far above the acceptance

levels.

[Insert Table 9]

Active ETFs vs. Mutual Funds

Recent rule developments present a challenge as to whether active ETFs could

substitute for more conventional mutual funds or comprise a complimentary solution.

One of the open questions is if actively managed ETFs will become an extra ETF share

class on the traditional open-end structure or a separate, innovative fund type. The

questions arise since the legislation of active ETFs is based on the existing regime of the

mutual funds.12

Gastineau (2003) argued that the addition of an ETF share class to a

conventional open-end fund has a positive effect on the existing shareholders‟ benefits

and a controversial effect on new shareholders. Moreover, “fund portfolio scales trades”

represent an offsetting positive factor, since the manager keeps the portfolio allocation

stable and limits the trades to inflows and outflows, keeping costs as low as possible.

Edelen et al (2007) assessed that the trading costs of the mutual funds are the highest,

proportionally, and the proportion depends on the capitalization scale of the fund. Zhao

(2002) argued that conventional mutual fund managers face the dilemma whether to

launch single or multi-class fund portfolios or to add one or more classes to the existing

fund. New classes in existing portfolios are primarily the results of the expansion of

traditional front-end loading and occur if the record track of the respective portfolio is

successful. Adverse front load funds have no reason to introduce a new share class13

. A

further obstacle for a conventional fund to add up an ETF share class in the existing

fund is that conventional mutual funds require a holding period to mature and cannot be

a rational solution for short term investors or traders. It is not clear if the multiple share

class approach will be a hazard for existing investors, with respect to tax treatment,

expenses and performance. In a traditional index fund the addition of extra share classes

is not a hazard for the current investors.

12

The replacements among others are attributed to the disclosure of the performance – NAV- as opposed to the benchmark and the disclosure of portfolio allocation. Conventional open-end funds are restricted to a quarterly disclosure within the next two months, which make up a 6-calendar-month horizon. Additional changes are based on the literature, the marketing material and the statutory limits. 13

Gastineau (2001) argued that an inherent constrain arises by the nature of ETFs, as the index based funds, such as S&P500 or Nasdaq 100, are useless when launched for 4 different shares classes, so the analysis is referring to more complex strategies.

14

Table 10, examines the ability of the active funds to outperform the conventional

mutual funds in terms of mean returns. The results of the unrestricted VAR indicate that

the pair PQY-Mutual funds exhibits mean reversion, however, the reversion is weaker in

the mutual funds. PQY also appears to be independent from the mutual funds. The

results of the pair PQZ-Mutual funds also exhibit mean reversion, even though the

reversion is weaker in the mutual funds. There is a negative interrelation from the mutual

funds to the active fund. The pair PSR-Mutual funds exploits mean reversion as well,

which is stronger in the mutual funds; however, there is significant negative feedback

from the active fund to the mutual funds. The results of the pair PMA-Mutual funds

demonstrate that the funds are totally independent, while, on the contrary, the mutual

funds are showing mean reversion and a positive relation feedback from the PMA.

Although the pair PLK-Mutual funds exhibits mean reversion, there is no other relation

between the pair.

[Insert Table 10]

In table 11, I compare the short and long run mean of the active and the passive

funds as indicated in (8) and (9) through a Wald test. The results indicate no difference in

excess return between the active ETFs and the mutual funds, as p-values range above the

acceptable levels.

[Insert Table 11]

Active ETFs vs. Hedge Funds

There are many challenging prospects that the hedge fund industry and

quantitative funds would face in an active ETF structure. The rule of daily disclosure and

the aim for more control and transparency on the part of the regulators, especially after

the subprime crisis, raises a crucial question that needs to be answered in the near future:

Are active ETFs the efficient and rational path for the quantitative hedge funds to be

regulated vehicles? Their nature, due to frequent rebalancing, has implicitly developed a

rational mechanism to monitor daily allocation performance while the daily disclosures

do not give rise to any additional costs. The second aspect is that of manager appraisal.

Managers in the hedge fund industry are already evaluated in a very strict and intensive

time horizon, since their performance is what constitutes the reputation of the fund. For

specific types of quantitative funds, recommended allocation comes out from the

optimizer, where there is no need for extended managers‟ comprehension. So, any type

15

of disclosure increases transparency and aids offshore funds in increasing their solvency

and their credibility to investors14

.

Table 12 represents the results of equations (14) and (15) between the returns of

the active funds and the returns of the hedge funds. The findings confirm positive excess

only for the hedge funds. Active ETFs exhibit a unidirectional link with the hedge funds

and their lagged values. The response that comes out from the hedge funds versus to the

active ETFs is almost two times higher than the response that comes out from their

lagged values. The hedge funds exhibit a significant positive relation with the active

ETFs. Between the two investment products the fit of the regression is greater in the

hedge funds.

Panel B performs a relative comparison between the short and long means of the

active ETFs and the hedge funds. The results indicate no difference in excess return

between the active ETFs and the hedge funds.

[Insert Table 12]

Concluding Remarks Active Exchange Traded Funds were under investigation in this paper. The

revolutionary decision of the SEC to permit trading of active ETFs ushered in a new era

in the asset management industry. However, law-mandated full transparency remains a

major obstacle.

Overall, the study proposes that although the infant period of the active funds is

not so successful, there is room for much improvement in the structure of the active

funds. Clearly, up to now, the first active ETFs have not yet been matched by asset

flows, while the first active ETFs can easy categorized as traditional equity index

strategies with limited added value for the investors. Of course, the notion of a broader

range of investment strategies could be more appealing to the investors. The next

launching of active ETFs includes strategies in short term fixed-income, money markets

and global macro. The results show that active ETFs are not as active as they appear to

the market participants. The link between active and passive ETFs is strong, but the

difference in performance and risk weighs on the side of the passive ETFs. Furthermore,

comparative evaluation with the mutual funds exhibits a superior performance and,

finally, there is a unidirectional relation between the active funds and the hedge funds,

since the former is influenced by the latter. Evidently, the results are subject to the low

14

There are exchange-traded funds (ETFs), known as quantitative based ETFs, which are supercharged by rules-based

and quantitative algorithms. The underlying index is based on a predefined algorithm.

16

activity and interest of the investors, since an increase in the popularity could improve

the performance of the active ETFs significantly.

The coming years will be enlightening as a number of industry sources say that the

many mutual fund companies that have made tentative steps to begin offering exchange

traded funds might leap first into the ETF business if the sec approves the exemptive

relief. According to industry sources large financial-services firms have filed for approval

to create ETFs. Clearly, the approval is a lengthy procedure. However, their empirical

evidence of an established customary track record of at least three years and the

launching of active ETFs from well-established investment managers will prove the

potential that can be achieved by the active ETFs.

17

References

Alexander, C., Barbosa, A. “Hedging index exchange traded funds” Journal of Banking

and Finance, 32, (2008), pp. 326-337.

Ackerman, C., McEnally, R., Ravenscraft, D. “The Performance of Hedge Funds: Risk,

Return and Incentive” Journal of Finance, 54 (1999), pp. 833-874.

Ascioglu, A., Aydogdu, M., Chou, R. K., Kugele, L., P. “An analysis of intraday pat-tern

in ETF and Common stock spreads.” Working paper, Bryant University, National

Central University and Christian Brothers University, 2006.

Boehmer, B., Boehmer, E. “Trading your neighbour‟s ETFs: Competition or

Fragmentation.” Journal of Banking and Finance, 27(2003), pp. 1667-1703.

Cabrera, J., Wang, T., Yang, J. “Short Horizon predictability of stock index ETFs: Does

Nonlinearity Help?” European Journal of Operational Research, 27 (2009), pp. 1627-

1703.

Cherry, J., “The Limits of Arbitrage: Evidence from Exchange Traded Funds.” working

paper, University of California-Berkeley, 2004.

Delcoure, N., Zhong, M. “On the premiums of Ishares.” Journal of Empirical Finance,

14 (2007), pp. 168-195.

Edelen, R., M., Evans, R., Kadlec, G. “Scale Effects in mutual funds performance: The

role of trading costs.” working paper, 2007.

Elton, E., Gruber, M., Comer, G., Li, K. “Spiders: Where are the Bugs?” Journal of

Business, 75 (2002), pp. 453-472.

Frino, A., Gallagher, D. “Tracking S&P 500 Index Funds.” Journal of Portfolio

Management, 28 (2001), pp. 44-55.

18

Frino, A., Gallagher, D., Neubert, A., Oetomo, T. “Index Design and Implications for

Index Tracking: Evidence from S&P 500 Index Funds.” Journal of Portfolio

Management, 30 (2004), pp. 89-95.

Gastineau, G. “Exchange-Traded Funds: An Introduction.” Journal of Portfolio

Management, 27 (2001), pp. 88-96.

-. “Converting Actively-Managed Mutual Funds to ETFs.” working paper, 2003.

-.“The Benchmark Index ETF Performance Problem.” Journal of Portfolio

Management, 30, (2004) pp. 96-104.

Gleasona, C., K., Mathurb, I., Mark, A. “Peterson Analysis of intraday herding behaviour

among the sector ETFs.” Journal of Empirical Finance, 11 (2004), pp. 681–694.

Goldsborough, R. “Active ETFs: A state of the Union” (2011), Morningstar, September

15.

Harper, J., T., Madura, J., Schnusenberg, O. “Performance Comparison Between

Exchange Traded Funds and Closed-End Country Funds.” Journal of International

Markets, Institutions and Money, 16 (2006), pp. 104-122.

Hasbrouck, Joel. “Intraday price formation in the market for U.S. equity indexes.” The Journal of Finance, 58 (2003), pp. 2285-2812.

Hegde, S., P., McDermott, J., B. “The market liquidity of Diamond, Qubes and their

underlying stocks.” Journal of Banking and Finance, 28 (2004), pp. 1043–1067.

Hendershott, T., Jones C., M. “Trade-through prohibitions and market quality.” Journal

of Financial Markets, 8 (2005), pp. 1-23.

Hughen, J., C., Mathew, P., G. “The efficiency of International Information Flow:

Evidence from the ETF and CEF Prices.” working paper, 2007.

19

Parkinson, M. “The extreme value method for estimating the variance of the rate of

return.” Journal of Business, 53 (1980), pp. 61–65.

Pennathur, A., K., Delcoure, N., Anderson, D. “Diversification benefits of iShares and

closed-end country funds.” Journal of Financial Research, 25 (2002), pp. 541–557.

Poterba, J., M., Shoven, J., B. “Exchange traded funds: a new investment option for

taxable investors.” American Economic Review, 92 (2002), pp. 422-427.

Rombotis, G., G., “Active vs. Passive Management: New Evidence from Exchange

Traded Funds.” working paper, 2009.

-. “Active versus Passive ETFs: An Investigation of Bid-Ask Spread.” The Journal of

Applied Finance, 16 (2010), pp. 5-25.

Tse, Y., Bandyopadhyay, P., Shen Y., P. “Intraday Price Discovery in the DJIA Index

Markets.” Journal of Business Finance & Accounting, 33 (2006), pp.572-1.585.

Tse, Y., Martinez, V. “Price Discovery and Informational Efficiency of International

iShares Funds.” Global Finance Journal, 18 (2007), pp. 1-15.

Zhao, X., Entry decisions by mutual fund families, Working Paper, College of William

and Mary, 2002.

20

Tables and Figures

Table 1 The table presents the seven features of the first 5 active ETFs and the mean, median and the standard deviation of the first four features. The sample period is extended by the inception of each ETF, such as April 16, 2008 for the first 4 ETFs and since November 24, 2008 for the PSR. The last observation has been taken on March 4, 2010.

PLK PMA PQY PQZ PSR Mean Median Std. Dev.

Annual Management Fee (%)

0,29 0,75 0,75 0,75 0,80 0,67 0,75 0,21

Expense Ratio (%) 0,29 0,75 0,75 0,75 0,80 0,67 0,75 0,21

Size ($ millions) 7,62 3,48 19,32 4,46 11,53 9,28 7,62 6,43

Daily Activity ($) 98.477 46.612 93.737 68.317 101.035 81.636 93.737 23.508

Age (months) 23 23 23 23 17 - - -

Style Bonds Equity-Large Cap

Equity-Large Cap

Equity-Multi Cap

Real Estate

- - -

Manager NO YES YES YES YES - - -

21

Table 2 The table presents OLS estimations of the first 5 active ETFs against three basic variables. Panel A presents the regression of the return of the active ETF. The sample period is extended by the inception of each ETF, such as April 16, 2008 for the first 4 ETFs and since November 24, 2008 for the PSR. The last observation has been taken on March 4, 2010. The age variable stands for the number of months counted by the inception of each ETF and the trading activity stands for a dummy variable taking the value of one if there is a daily trading of the respectively ETF. The corresponding p-values are reported for each separate variable and statistics are corrected for autocorrelation and heteroscedasticity using the Newey-West estimator with 5 lags.

PLK PMA PQY PQZ PSR

Intercept 0,000 -0,002 -0,002 -0,004 0,001

0,805 0,081 0,041 0,050 0,615

AGE 0,000 0,001 0,001 0,001 -0,001

0,805 0,082 0,045 0,053 0,619

Trading activity 0,001 0,022 -0,011 -0,012 0,343

0,965 0,823 0,907 0,914 0,418

Turnover 0,000 0,000 0,000 0,000 0,000

0,942 0,000 0,219 0,920 0,766

R2 0,000 0,008 0,005 0,004 0,002

Observations 474 474 474 474 319

22

Table 3 The table presents OLS estimations of 20 days moving correlation of the active ETFs. The sample period is extended by the inception of each ETF, such as May 13, 2008 for the first 4 ETFs, since December 24 for the PSR. The last observation has been taken on March 4, 2010. Dummy variable stands for an indicator for lagged negative return of benchmark and the cross term is the product of the dummy variable and the lagged correlation. The corresponding p-values are reported for each separate variable.

Return PLK PMA PQY PQZ PSR

Intercept 0,006 0,021 0,028 0,020 0,010

0,365 0,088 0,015 0,103 0,566

Dummy Variable 0,001 0,006 -0,019 0,016 0,003

0,895 0,729 0,300 0,364 0,878

Lagged Correlation

0,953 0,965 0,966 0,978 0,988

0,000 0,000 0,000 0,000 0,000

Cross Term -0,017 -0,004 0,010 -0,033 -0,008

0,627 0,874 0,681 0,177 0,721

R2 0,894 0,924 0,933 0,932 0,972

Observations 455 455 455 455 299

23

Table 4 Panel A represents the descriptive statistics in percentage for the first active ETFs. The sample periods have been split up to six, twelve, and eighteen months, and at last I represent the estimations for the entire sample period in order to check any abnormalities as the funds are growing up. Panel B represents the tracking error between the active ETF and the respective passive ETF. The sample period is extended by the inception of each ETF, so since April 16, 2008 for the first 4 ETFs, and November 24, 2008 for the PSR and ends on March 4, 2010. The n/a notation indicates for the concrete period the respective ETF have not been launched.

ACTIVE PASSIVE ACTIVE PASSIVE ACTIVE PASSIVE ACTIVE PASSIVE ACTIVE PASSIVE

Panel A Sample

Period In Months

PLK SHY PMA SPY PQY QQQQ PQZ SPY PSR FTY

Mean 6 0,002 0,009 -0,311 -0,264 -0,248 -0,242 -0,367 -0,264 n/a n/a

12 0,000 0,011 -0,126 -0,166 -0,156 -0,108 -0,237 -0,166 0,204 0,261

18 -0,002 0,009 -0,033 -0,047 -0,038 -0,004 -0,098 -0,047 0,225 0,238

All 0,001 0,009 -0,017 -0,029 -0,009 0,011 -0,069 -0,029 0,203 0,189

Median 6 0,000 0,025 0,000 0,043 0,000 -0,188 -0,162 0,043 n/a n/a

12 0,000 0,025 0,000 0,040 0,000 -0,142 0,000 0,040 0,000 0,585

18 0,000 0,024 0,000 0,064 0,000 0,064 0,076 0,064 0,000 0,374

All 0,000 0,013 0,000 0,108 0,000 0,103 0,074 0,108 0,000 0,319

Maximum 6 2,823 0,709 7,085 13,560 12,726 11,452 13,032 13,560 n/a n/a

12 2,922 0,709 9,473 13,560 12,726 11,452 13,032 13,560 15,932 22,602

18 2,922 0,709 9,473 13,560 12,726 11,452 13,032 13,560 15,932 22,602

All 2,922 0,709 9,473 13,560 12,726 11,452 13,032 13,560 15,932 22,602

Minimum 6 -2,598 -0,560 -9,626 -10,364 -10,286 -9,396 -12,267 -10,364 n/a n/a

12 -3,001 -0,560 -9,626 -10,364 -10,995 -9,396 -13,210 -10,364 -18,874 -19,089

18 -3,001 -0,656 -9,626 -10,364 -10,995 -9,396 -13,210 -10,364 -18,874 -19,089

All -3,001 -0,656 -9,626 -10,364 -10,995 -9,396 -13,210 -10,364 -18,874 -19,089

Std. Dev. 6 0,641 0,193 1,943 2,427 2,593 2,407 2,803 2,427 n/a n/a

12 0,731 0,169 2,412 2,819 2,661 2,666 3,411 2,819 4,569 6,708

18 0,665 0,153 2,089 2,418 2,326 2,288 2,946 2,418 3,993 5,080

All 0,622 0,142 1,918 2,210 2,141 2,104 2,686 2,210 3,436 4,351

Sharpe Ratio

6 0,003 0,048 -0,160 -0,109 -0,096 -0,101 -0,131 -0,109 n/a n/a

12 0,000 0,067 -0,052 -0,059 -0,059 -0,040 -0,069 -0,059 0,045 0,039

18 -0,003 0,059 -0,016 -0,019 -0,017 -0,002 -0,033 -0,019 0,056 0,047

All 0,002 0,062 -0,009 -0,013 -0,004 0,005 -0,026 -0,013 0,059 0,043

Stewness 6 0,253 -0,018 -1,037 0,360 0,133 -0,007 -0,337 0,360 n/a n/a

12 0,005 -0,115 -0,244 0,239 0,104 0,191 -0,517 0,239 0,073 0,049

18 -0,096 -0,311 -0,342 0,124 -0,029 0,084 -0,677 0,124 -0,038 0,001

All -0,094 -0,313 -0,379 0,100 -0,075 0,051 -0,753 0,100 -0,011 0,024

Kurtosis 6 9,981 4,772 10,236 12,936 9,163 8,524 10,012 12,936 n/a n/a

12 8,081 5,020 6,750 6,633 8,413 5,406 6,223 6,633 7,602 3,864

18 8,413 6,089 8,159 8,172 9,687 6,676 7,652 8,172 7,103 5,505

All 8,978 6,611 9,253 9,393 10,877 7,521 8,925 9,393 9,134 7,220

Panel B: Tracking Error (%) Active vs. Passive ETF

0,38 1,19 1,01 1,13 2,37

24

25

Table 6

The table presents the pairwise mean and median correlation between the active and passive ETFs. The estimations have been considered for three different horizons, 5-days, 20-days and 60-days. The sample period is extended by the inception of each ETF, such as April 16, 2008 for the first 4 active ETFs and since November 24 for the PSR. The last observation has been taken on March 4, 2010.

PLK-SHY PMA-SPY PQY-QQQQ

5days 20days 60days 5days 20days 60days 5days 20days 60days

mean 0,082 0,115 0,129 0,579 0,617 0,619 0,690 0,708 0,729

median 0,040 0,121 0,149 0,739 0,651 0,656 0,825 0,762 0,739

PSR - FTY PQZ-SPY

5days 20days 60days 5days 20days 60days

mean 0,591 0,629 0,633 0,637 0,693 0,716

median 0,786 0,777 0,641 0,778 0,751 0,733

26

Table 7

The table represents the empirical estimations of the active and passive ETFs. The definition of the active ETFs and the data span is analysed in table (1). In Panel A and B, I run an OLS regression against the daily return and the daily risk of each ETF respectively. The risk stands for the range-based estimators. The table represents only the factors that found to be significant. All table enters multiplied by one hundred. The corresponding p-values are reported for each separate variable and statistics are corrected for autocorrelation and heteroscedasticity using Newey-West estimator with 5 lags.

ACTIVE PASSIVE ACTIVE PASSIVE ACTIVE PASSIVE ACTIVE PASSIVE

Panel A: Return PMA SPY PQY QQQQ PQZ SPY PSR FTY

Intercept -0,001 -0,001 0,004 -0,001 0,002 -0,001 0,002 0,001

0,270 0,040 0,000 0,081 0,040 0,040 0,112 0,360

Drussell Value - Growth

0,347 0,579

0,579 1,627 2,682

0,075 0,000

0,000 0,000 0,000

HEDGE FUND INDEX

1,649 1,525 1,933 1,279 3,055 1,525

0,000 0,000 0,000 0,001 0,000 0,000

Drussell Large - Small

-0,271

-0,509 -0,436 -0,271 -1,102 -1,781

0,014

0,000 0,017 0,014 0,000 0,000

Volatility Momentum

-0,082 -0,192 -0,131 -0,188 -0,160 -0,192

-0,176

0,000 0,000 0,000 0,000 0,000 0,000

0,000

Volume 0,003 0,003 -0,006 0,004 -0,004 0,003

0,062 0,004 0,000 0,001 0,021 0,004

R2 0,305 0,726 0,412 0,633 0,508 0,726 0,215 0,572

Observations 474 474 474 319

Panel B: Volatility

PMA SPY PQY QQQQ PQZ SPY PSR FTY

Intercept 0,000 0,000 0,000 0,000 0,000 0,000 0,000 0,000

0,002 0,000 0,002 0,000 0,004 0,000 0,000 0,000

Drussell Value - Growth

0,015

0,074

HEDGE FUND INDEX

-0,014 -0,027

-0,031

-0,027

0,023 0,002

0,000

0,002

Drussell Large - Small

-0,011 -0,012

0,038 0,044

Volatility Momentum

0,001

0,091

Volume 0,000 0,000 0,000 0,000 0,000 0,000 0,000

0,074 0,072 0,003 0,024 0,071 0,072 0,003

R2 0,018 0,086 0,018 0,153 0,038 0,086 0,058 0,022

27

Observations 474 474 474 319

Table 8

The table represents the results of an unrestricted VAR specification between the daily mean of the active ETF versus the daily mean of the passive ETF. The respective index has been applied as exogenous variable. The definition of the active ETFs and the data span is analysed in table (1). The t-statistic is represented in brackets.

Active ETF

Passive ETF

Active ETF

Passive ETF

Active ETF

Passive ETF

PQY QQQQ PQZ SPY PSR FTY

C -0,008 -0,008 C -0,001 0,000 C 0,001 -0,001

[-1,53] [-1,37] [-1,00] [-1,52] [ 0,62] [-1,13]

Active ETF PQY(-1) -0,493 -0,116 PQZ(-

1) -0,299 0,022 PSR(-1) -0,113 -0,004

[-7,72] [-1,75] [-7,14] [ 2,44] [-2,03] [-0,14]

Passive ETF QQQQ(-

1) 0,396 -0,027 SPY(-1) 0,433 0,006 FTY(-1) 0,060 0,052

[ 6,09] [-0,40] [ 8,46] [ 0,59] [ 1,29] [ 2,19]

Benchmark NASDAQ 0,000 0,000 S&P500 0,976 0,992 FTSE

NAREIT 0,374 0,886

[ 1,52] [ 1,41] [ 29,70] [ 142,12] [ 9,20] [ 43,00]

R-squared

0,116 0,023 0,664 0,978 0,233 0,866

Observations

474 474 318

Active ETF

Passive ETF

Active ETF

Passive ETF

PMA SPY PLK SHY

C 0,000 0,000 C 0,000 0,000

[-0,09] [-1,49] [-0,18] [-0,02]

Active ETF PMA(-1) -0,299 -0,001 PLK(-1) -0,389 0,006

[-7,36] [-0,056] [-9,13] [ 1,20]

Passive ETF SPY(-1) 0,442 0,033 SHY(-1) 0,136 -0,124

[ 12,42] [ 3,80] [ 0,73] [-5,51]

Benchmark Russell

200 0,540 1,018

Barclays 1-3 years

0,482 0,883

[ 17,89] [ 139,49] [ 2,55] [ 38,63]

R-squared 0,472 0,977 0,156 0,765

Observations 474 474

28

Table 9

The table represents the results of Wald test between the daily mean of the active ETF versus the daily mean of the passive ETF. The definition of the active ETFs and the data span is analysed in table (1). The p-values are reported in nominal form.

Active ETF

Passive ETF

Active ETF

Passive ETF

Active ETF

Passive ETF

Active ETF

Passive ETF

Active ETF

Passive ETF

PQY QQQQ PQZ SPY PSR FTY PMA SPY PLK SHY

Hypothesis Ho: Intercepts are equal

X2 0,006 X2 0,442 X2 1,133 X2 0,068 X2 0,030

p-value 0,9361 p-value 0,506 p-value 0,287 p-value 0,794 p-value 0,863

Hypothesis Ho: Long term means are equal

X2 0,084 X2 0,312 X2 1,235 X2 0,139 X2 0,029

p-value 0,772 p-value 0,576 p-value 0,267 p-value 0,710 p-value 0,864

29

Table 10

The table represents the results of an unrestricted VAR specification between the daily mean of the active ETF versus the daily mean of the mutual funds. The definition of the active ETFs and the data span is analysed in table (1). The t-statistic is represented in brackets.

Active ETF

Mutual Funds

Active ETF

Mutual Funds

Active ETF

Mutual Funds

PQY PQZ PSR

C 0,000 0,000 C -0,001 0,000 C 0,002 0,000

[-0,11] [-0,28] [-0,66] [-0,44] [ 1,14] [ 0,08]

Active ETF PQY(-1) -0,205 0,005 PQZ(-1) -0,110 -0,061 PSR(-1) -0,143 0,003

[-4,54] [ 0,11] [-2,42] [-1,73] [-2,56] [ 0,04]

Mutual Funds

Mutual Funds (-1)

-0,012 -0,118 Mutual

Funds (-1) -0,126 -0,068

Mutual Funds (-1)

-0,065 -0,273

[-0,25] [-2,58] [-2,13] [-1,48] [-1,63] [-5,02]

R-squared

0,042 0,014 0,021 0,011 0,027 0,074

Observations

474 474 318

Active ETF

Mutual Funds

Active ETF

Mutual Funds

PMA PLK

C 0,000 0,000 C 0,000 0,000

[-0,20] [-0,23] [ 0,08] [ 1,78]

Active ETF PMA(-1) -0,045 0,134 PLK(-1) -0,379 0,015

[-0,97] [ 2,90] [-8,89] [ 1,22]

Mutual Funds

Mutual Funds (-1)

0,000 -0,135 Mutual

Funds (-1) -0,013 0,132

[ 0,01] [-2,97] [-0,087] [ 2,90]

R-squared 0,002 0,038 0,144 0,020

Observations 474 474

30

Table 11

The table represents the results of Wald test between the daily mean of the active ETF versus the daily mean of the mutual funds. The definition of the active ETFs and the data span is analysed in table (1). The p-values are reported in nominal form.

Active ETF

Mutual Funds

Active ETF

Mutual Funds

Active ETF

Mutual Funds

Active ETF

Mutual Funds

Active ETF

Mutual Funds

PQY PQZ PSR PMA PLK

Hypothesis Ho: Intercepts are equal

X2 0,013 X2 0,064 X2 0,376 X2 0,000 X2 0,183

p-value 0,909 p-value 0,801 p-value 0,540 p-value 0,982 p-value 0,669

Hypothesis Ho: Long term means are equal

X2 0,016 X2 0,057 X2 0,440 X2 0,000 X2 0,469

p-value 0,901 p-value 0,812 p-value 0,507 p-value 0,992 p-value 0,493

31

Table 12

The table represents the results of a cross sectional estimation between the return of the active ETFs versus the mean return of the group of Hedge Funds. The definition of the active ETFs and the data span is analysed in table (1). Panel B, represents the results of Wald test between the means of the active ETFs versus the hedge funds. The corresponding p-values are reported for each separate variable.

Panel A Active ETFs Hedge Funds

C -0,014 0,010

0,042 0,033

Active ETFs - 0,795

- 0,000

Active ETFs (-1) 0,284 -0,140

0,069 0,307

Hedge Funds 0,644 -

0,000 -

Hedge Funds (-1) -0,076 0,164

0,624 0,349

R-squared 0,632 0,869

Observations 53

Panel B Hypothesis Ho: Intercepts are equal

X2 1,175 6,372

p-value 0,001 0,012

32

Figure 1: The figure represents the 5days moving average correlations between the

active (PQZ) and the passive ETF (QQQQ) at the left axis. At the right axis the figure

represents the Nasdaq 100 index. The sample period is since April 16, 2010 and extended

up to March 4, 2010.