Problems with data and why these data, if incomplete could lead to misleading
conclusions.
This is a brief note on the problems of collecting and analysing data on the numbers and
purposes of discretionary trusts.
A dataset of trusts based offshore would presumably include information such as the
number of trusts, what kind of trust they are eg. discretionary, whether there are powers
attached, this data set would not include the value of the trust fund. Even if it did such a
dataset would provide no useful information on whether the trust was established for the
purpose of avoiding the requirements of the Savings Directive.
As there is no dataset in existence from prior to the implementation of the Directive it
would not yield any information as to whether any of the trusts may have been
established for the above purpose.
Therefore, policy analysts would need to explore other ways to discover whether
discretionary trusts had been misused in such a way.
MANOVA analysis of data
A simple linear regression analysis would be possible (assuming that the data is normal
which it might not be given the differences in places where you might put a trust – there
are endogenous factors standard across certain datagroups – different jurisdictions - as
opposed to the entire dataset). However, it would be necessary to test multiple variables.
There are likely to be multiple influences on the decision to circumvent the provisions of
the directive. You may establish multiple variables eg. price, price relative to other
products, quality of service, political stability, quality of advisors. There are demand side
factors also; demand elasticity of price, price relative to other products demand elasticity
of political stability, the creation of the savings directive (which could also be measured
as an elasticity measurement, if the data were available). That means you need to
establish a multiple regression model.
However you may also have more than one dependent variable for example if you also
factor in the establishment of companies. To represent such mutually or jointly dependent
or endogenous variables you need to create simultaneous-equation model. However,
because of the interdependence of the endogenous variables it is inappropriate to apply
the OLS method to an individual equation in the system. The process of developing
reliable multiple regression models is highly complicated, even assuming that data are
available. Using such a model it would be possible to ascertain the level of use of trusts
as a vehicle for circumventing the Savings Directive.
However, as logicians and statisticians are more than aware probability of a type 2 error
increases as that of a type 1 error decreases. ie. When you reduce the risk of rejecting the
true hypothesis you increase the probability of accepting a false hypothesis.
There is a tradeoff therefore between these two types of errors. The only way in which it
is possible to balance that tradeoff is to find out the relative costs of the two types of
errors. Then, if the error of rejecting the null hypothesis which is in fact true (Error Type
1) is costly relative to the error of not rejecting the null hypothesis which is in fact false
(Error Type II), it will be rational to set the probability of the first kind of error low. If, on
the other hand, the cost of making Error Type 1 is low relative to the cost of making
Error Type II, it will pay to make the probability of the first kind of error high. Of course,
this problem can be circumvented if it is possible to gain a probability value. This whole
problem is extremely important given the likely small sample size of any data.
Nonetheless, to build such a model data about numbers of trusts, the customer base, the
costs of administering them, and other significant factors in terms of costs and
opportunity costs are simply never going to be available. Most importantly a figure of the
number of trusts in a jurisdiction is not an indicator of their being ‘a problem’.
I suggest a more simple way of carrying out this research – namely conducting a mystery
shopping exercise amongst institutions where you consider that there is a problem and
see what they tell you.
JR 25th September 2007