Performance metrics for Spanish investment funds

Luis Ferruz

*

, Christian Pedersen and Jose

´

L. Sarto

*

Department of Accounting and Finance, Faculty of Economics and Business Studies,

University of Zaragoza, C/ Gran Vı

´

a 2, Zaragoza 50005, Spain.

E-mail: [email protected]

´

Zaragoza, Spain.

Practical applications

This paper is useful for investors and managers of investment funds since it tries to identify the true

performance of these portfolios. The empirical results obtained in the Spanish equity fund market

provide evidence for the incorrect performance valuations when classical indexes are considered.

These problems are caused by: 1) the asymmetric return distributions; 2) the negative return premia.

On this subject, the application of the original measures proposed in this work leads to coherent

performance rankings. So, these rankings suppose very useful information for fund investors to value

the management quality of each fund and for fund managers to know their relative position with

time period studied. We apply alternative risk

measures, such as semi-standard deviation and absolute

deviation and test if the associated performance

measures provide markedly different rankings from the

classic indices. We find that one subset of funds

analysed displays negative return premia, and make an

additional adjustment to the suggested performance

metrics. Overall when comparing the rankings, we see

Derivatives Use, Trading & Regulation Volume 12 Number 3 2006 219

important recent phenomena in Spanish

financial markets. The growth of the assets under

management by Spanish funds have been one of

the biggest in Europe for the past 15 years, with

a compounded annual rate of growth greater

than 25 per cent. In all, 250bn euros are

currently managed by approximately 2,600

Spanish investing funds.

Investment funds are now the third most

important investment alternative in Spanish

majority of small funds. The overall picture

points to a set of vehicles reflecting illiquidity

and ‘youth’ — often leading to strong

asymmetric return distributions.

In this study, we examine the degree of

asymmetry of the returns of these funds and then

evaluate different metrics for capturing their

performance. To determine the performance

values for each fund, we initially consider three

of the indices most commonly implemented in

In line with studies by Damant et al.,

Eftekhari et al.,

Okunev

and Pedersen and Satchell,

variance, and therefore standard deviation, to

measure the total risk in portfolios was not the

most suitable method when the historical return

distributions of said portfolios display problems

and propose alternative ratios for performance

— this was taken further in Eftekhari et al.

who examined such

alternative metrics in more detail. Hence, these

are also tested to see if different from the

traditional performance metrics.

Such alternative measures have also appeared in

the works of Damant and Satchell,

Hwang and

Pedersen,

Campbell and

Kraeussl,

Cheng,

Gu,

Post and Van Vliet

and

Morton et al.

Our preliminary examinations also reveal that

a subset of those portfolios considered display a

negative return premium in comparison to the

risk-free assets considered. Hence, we apply

Finally, we examine the levels of correlation

existing among all the performance indices taken

into consideration to opine on whether moving

to the less traditional metrics is justified

empirically as well as theoretically — that is we

will see whether rankings of real fund

performance changes significantly.

The fundamentally differential elements of

this work are its focus on a relatively infrequently

analysed market, as is the case with the Spanish

220 Ferruz, Pedersen and Sarto

supplement these with semi-standard deviation

and absolute deviation (to capture asymmetry).

Additionally, both these ‘new’ metrics and the

Sharpe Ratio with the above adjustments in

consideration of the return premium in a relative

sense as opposed to in absolute terms. Hence, we

have six metrics tested in all.

Specifically, in accordance with these

considerations, in addition to the S

Ratio. The first of these is expressed as follows:

p

SSD

p

where E

p

represents the average return on

portfolio p; R

f

index, although the return premium is expressed

in relative terms. The semi-standard deviation is

calculated as follows:

SSD

p

1

n

n

ðmin½0; R

pt

p

AD

p

p

calculated as follows:

AD

p

p

p

immediately apparent. Finally, Table 1 indicates the

As can be seen, the correlation between the

measures is very high. What this implies is that

although the theory suggests we change from

traditional to asymmetric measures corrected for

the negative excess returns, in practice there is

not a material change in performance rankings

of the funds from doing so.

traditional measures of performance (Sharpe,

Treynor and Jensen) are applicable. The study

Table 1: Correlation coefficients between two-by-two performance rankings

S

222 Ferruz, Pedersen and Sarto

return distributions in 74 of the 85 portfolios

analysed — this represents a problem when

employing common total risk measures such as

variance or standard deviation, which require

symmetric return distributions to be valid for

general investor preferences. Moreover, we

observe that a subset of the investment funds in

the sample do not attain the average return

corresponding to risk-free assets, a situation

which on the one hand, include risk

measurements such as semi-standard deviation or

absolute deviation and, on the other, approach

the return premium in a relative sense.

from the application of all the measurements

considered, however, barely differ from one

another. What this suggests is that despite the

anomalies described above, the traditional

performance metrics to be applied. While this

DISCLAIMER

We stress that the opinions stated in this paper

exclusively reflect the view of the author and are

not those of Mercer Oliver Wyman.

224 Ferruz, Pedersen and Sarto

226 Ferruz, Pedersen and Sarto