Study of pension plans;Use of statistical sampling rises;

              A STUDY OF PENSION PLANS
A gap in the existing literature on pen­sions
has been filled by publication in
January of the Firm's study, The Pen­sion
System in the United States. Prin­cipals
Robert L. Potter (Cincinnati) and
John C. Tillotson (Executive Office) did
the research and the original writing on
which the study is based. Executive
Office partners Paul W. Pinkerton and
Everett J. Shifflett, respectively, edited
and directed the work.
Private pension plans have had a
tremendous growth in the past two dec­ades.
Together with Social Security
contributions, costs of private pension
plans have become a significant item of
business costs and liabilities. This new
The year 1958 marked the start of the
Firm's intensive investigation, devel­opment,
and testing of statistical sam­pling
as a new audit tool. Kenneth W.
Stringer, Executive Office, the partner
who had most to do with the project,
was helped in the later stages by a com­mittee
of five partners and by many
other people from the practice offices.
The result is the Sampling Plan, pub­lished
in October 1962 as Technical
Procedure Manual Bulletin 301.
New Mathematical Concepts
During those four years H & S devel­oped
some entirely new mathematical
concepts, in cooperation with Professor
Frederick F. Stephan of Princeton, and
devised a method of relating statistical
sampling to internal control.
Now, six years after this work began,
we have encouraging reports on the
Plan's application. Comparing the
Firm's use of statistical sampling dur­ing
the busy season 1961-62 (while the
Plan was in its testing stage) and 1962-
63, we can see a marked increase.
booklet is not intended to make the
reader an expert on pension plans; what
it does is to bring basic information to­gether
in one place in a form that is
readily usable by our people who meet
with pension plans in their day-to-day
work. Copies have been provided to all
professional personnel of the Firm.
Study Contents
The booklet begins with a look at
governmental systems — mainly Social
Security, which now covers some 85
million people. It then describes private
plans, both insured and trusteed, the
types of benefits and the ways they are
provided, and various methods of fi-
There are indications, too, that it will
be used much more in the future. Re­ports
from the offices comparing these
two periods show that the number of
engagements in which the Sampling
Plan was used more than doubled, from
150 to 358. The total number of appli­cations
of the Plan increased from 389
to 886.
Statistical sampling affords more ob­jective
choices of how much informa­tion
to test and better means for eval­uating
errors uncovered by the tests.
The Sampling Plan measures and con­trols
the risks inherent in testing prac­tices.
There is reason to believe it re­sults
in a more uniform application of
auditing judgment throughout the
Firm.
Experience with the Plan shows that
it does not change total audit time
greatly. In 96 per cent of recent cases
in which it was used, total audit time
did not change by more than 5 per cent.
The Plan is being used effectively for
client engagements of all sizes.
When Bulletin 301 was written, it
nancing them. One chapter assumes an
employee group with stated benefits
and then compares costs and fund bal­ances
under several funding methods.
Another chapter outlines federal in­come
tax aspects of pension plans. A
bibliography is provided.
Besides our own people, others who
may find the study helpful are clients,
bankers, and lawyers who have felt the
need for a general understanding of
pension plans. Smaller companies that
do not yet have pension plans will find
it useful as an introduction to the sub­ject;
and companies that already have
plans may note alternative methods of
financing them.
was thought that three stratums were a
practical maximum in sample design.
Our experience now shows that multi-stratum
design (often four or more stra­tums,
and sometimes up to twelve and
even twenty) is feasible and results in
substantial reduction in sample size.
The great majority of applications to
tests of transactions have covered a
full year. This reflects recognition of
the fact that while a sample selected
from the transactions of one month
gives assurance only with respect to
that month, in most cases only moder­ately
greater samples give assurance as
to the transactions of an entire year.
These encouraging results can be at­tributed
also to increased familiarity
with the Plan and more efficient selec­tion
techniques, including increased
use of clients' data processing equip­ment
in the Plan's application.The Firm
is now considering the practicability of
a standard EDP program for sample
design which our auditors would be
able to apply in all cases on clients'
equipment.
USE OF STATISTICAL SAMPLING RISES
7