20(34)
Data
Data for the analysis were collected from procurement operations in central
and local governments during 2007 and 2008. In all, the sample comprises
76 procurement operations from central government and 50 operations
from a municipality. In the central government sample, all operations
supported by the Agency for Support to Higher Education (VHS) are
included. This agency has worked as a consultant to other central
government agencies, and the sample consequently comprises goods and
services from a wide variety of sectors. The local government sample was
collected from the municipality of Sunne in western Sweden, a typical
Swedish municipality with 13 600 inhabitants, close to the median. General
descriptive statistics are given in Table 1.
Table 1.
General Statistics for the Data Sample
Central
government
Local
government
Total number of operations
76
50
Average number of tenders per procurement
operation
5.26
4.82
Number of operations based on price only
13
13
Number of operations with only 1 tender
14
5
This sample may seem to be somewhat thin, but for policy purposes—
determining a reasonable threshold value—only the distribution of relative
gains is necessary. This distribution depends on three parameters—mean,
variance, and the proportion of operations with only one tender—and for
estimating these three, the sample is sufficient in size.
An example of the distribution functions (supply curves) that have been
generated from the data is shown in Figure 1. Once the distribution
function has been estimated, it is possible to compute the expected price
if, for instance, three randomly selected suppliers are invited. Given the
average number of tenders in open procurement (5), it is also possible
to compute the expected price for that situation. The gain, as defined
previously, is simply the difference between the average price level and the
expected price in the procurement situation.
