18(34)
An important observation that indicates a feasible solution is that
procurement operations fall into two distinctive classes: one in which
quality dimensions are important, being weighted by at least 50 percent,
and a second category in which quality is only used for marginal
adjustments of the ranking based on price (quality weight at most 20
percent).
The criterion suggested as a solution for the former category is to use the
ratio between quality and price for the alternatives under comparison. This
is simple to use and is supported by the text of the relevant EU directive
(―best value for money‖; Dir. 2004/18/EC, pt. 46 of the preamble). If the
procurer wishes to eliminate the risk that low-quality alternatives win the
contest by offering very low prices, restrictions on the lowest acceptable
quality should be entered into the terms of reference. Such restrictions if
used should not affect the zero level of the quality scale used by the
procuring entity.
For the second category, where the procuring entity has signaled that price
is far more important than quality, ranking on price alone has been used.
It has been verified that the ranking of alternatives in these cases was not
altered as a result of this modification.
In line with this criterion, the independent variable used in the present
study for ranking suppliers and forming supply curves when quality aspects
are important in the evaluation is the
price per quality unit
. When quality
is less important, the independent variable is
price
alone, as in the cases
where only price was used in the original evaluation.
Gains
Recall that the
expected value
over the population of potential tenders is
taken as the reference price in the present study. This can be thought of as
the outcome that would result if the procuring entity picked a producer at
random among those prepared to deliver the goods or services demanded.
Following this definition, the gain from an actual procurement operation
is computed in the following way. Based on the tenders submitted, a
distribution function is estimated by using a lognormal function with three
parameters: the zero value (
τ
), the average (μ), and the standard deviation
(σ). The gain is the difference between the expected value and the lowest
price among the tenders. If the population of procurements is normalized,
the average gain will approach the gain computed from order statistics by
using the average number of tenders in the population.
By using the distribution function thus derived, it is also possible to answer
questions of the type, ―What would happen to the price level if, instead of
carrying out a full-scale procurement, the procuring entity approaches, say,
three suppliers chosen at random from the population of suppliers?‖
The gain from procurement computed in this way is an underestimate, for
at least two reasons. Firstly, the effect of merely subjecting suppliers to
potential competition is not included. If only one tender is submitted, the
gain will be zero by definition. This underestimates the real effect compared
with the situation in which the procuring entity approaches a producer
directly, as the following example shows. In one of the procurement
