Abstract :
[en] Evaluation of the performance of investment funds
is a topic of considerable interest to practitioners
and academic researchers. Performance indicators
of fi nancial places have for long posed interesting
challenges with regards to funds investors, but also to
legal regulators authorities. The two major issues that
need to be addressed in any performance study are how
to choose an appropriate benchmark for comparison and
how to adjust a fund’s return for risk. Indeed, investors
desire information about representative market indexes
as a norm to evaluate the performance of their portfolios.
MSCI Indexes are frequently used by institutional investors
around the world as benchmarks to decide allocation
of funds across asset classes and regions. Despite this
wide acceptance, MSCI country index has in its original
application a number of drawbacks and limitations. The
main problems can be traced to the presence of usual
biases, such as sampling, survivorship and instant history
biases (Fung and Shieh, 2002), involving problems
into the aggregation procedure. Thus, one can explain
why certain fi nancial places are less representative, specifi
cally for funds distribution places. Some results also
indicate problems related to misclassifi cation in mutual
funds (Sharpe, 1992). Each of these phenomena can have
a signifi cant impact on international diversifi cation for
fund managers. Based on these empirical fi ndings, Ferreira,
Miguel and Ramos (2007) examine cross-country
mutual fund performance using several alternative benchmark
models including a domestic and an international
version of the Carhart (1997) four-factor model. Using
multiple regressions, they obtain signifi cant determinants
explaining funds performance, like the funds size, the
fees, the management style… .
In this article, we contribute to the existing discussion
on alternative benchmarks to compare fi nancial places,
by conducting an analysis on funds time variation
structure. Contrary to previous literature, we propose
to use directly the information contained in the NAV to
extract performance characteristics of funds. Then, each
domiciliation place is compared by constructing a fund
synthetic index that will capture the time structure of
mutual fund performance.
Usual statistical approach consists to estimate fi nancial
returns of each fund in each country, which involves dealing
with huge data sets that may cause the calculation processes
to become slow and cumbersome and the results
diffi cult to be interpreted and used in further applications.
To reduce data sets, and give conclusions for each fi nancial
place, indicators of the mean of fund’s returns can
be used. Then one can identify classes of domicile funds
that are subject to common properties. But this classical
approach, gives only approximate results because it is based
on an aggregation of average performance and risk and a
boxplot statistical format. The construction of fund synthetic
portfolio avoids this issue. First, it avoids the logic of
representativity through market capitalization that is often
diffi cult to apply to the mutual fund universe. Second, it is
based on factor analysis techniques to generate indexes that
are able to capture a very large fraction of the information.
More precisely, it permits to take into account the common
properties of fund returns relative to their domicile while
keeping the maximum of information given by the original
data. Indeed, it may be very useful to use a transformation
to form a simplifi ed data set retaining the characteristics of
the original data set. Principal component analysis, abbreviated
as PCA, is a method of statistical analysis useful in
data reduction and interpretation of multivariate data sets
by identifying factors of common behavior such that not
much of the contained information is lost. In our context,
we use this method to derive portfolio weights in order to
construct a synthetic portfolio of each fi nancial place. For
that, we propose to replace the matrix of returns and to
derive an index which keeps the global representation of
each fi nancial place.