25/03/2019
Proximus v Council of the European Union [T-117/17]
In previous articles we have written about the importance of bidders and contracting authorities robustly testing the formulae used to assess price. A recent decision of the EU General Court serves to highlight the practical importance for bidders of understanding the way that the authority’s price methodology will work and ensuring that their bid is tailored appropriately to give the best possible score. It also shows some of the difficulties for contracting authorities of getting their financial evaluation right and highlights the importance of robustly testing financial assessment formulae to ensure they will deliver an accurate and defensible output.
Background
The European Council sought tenders for the provision of a framework contract for the supply of cybersecurity services. Several bidders responded, including the claimant, Proximus. The contract was to be awarded on the basis of a 60/40 quality/price split.
Financial evaluation
The services were split into four ‘service packages’ corresponding to the different types of services required under the framework (though the procurement was not divided into Lots). Bidders were required to provide a price for each service package. The financial evaluation compared bidders’ price per service package to the sum of all bids for that service package with the lowest price bid gaining the most points.
The example below shows, at a basic level how the formula worked, and one of the ways in which it could be manipulated:
|
|
Service package 1 prices |
Service package 2 prices |
Service package 3 prices |
Service package 4 prices |
Bidder’s total prices |
|
Bidder 1 |
100 |
100 |
100 |
100 |
400 |
|
Bidder 2 |
80 |
90 |
120 |
120 |
410 |
|
Bidder 3 |
295 |
5 |
195 |
5 |
500 |
|
Total per service package |
475 |
195 |
415 |
225 |
1310 |
We can see that Bidder 1 has chosen to split their price roughly equally across all service packages. Bidder 2 has split in different proportions but has bid a very similar overall price to Bidder 1 whilst Bidder 3 is the most expensive but has loaded two of the service packages with high prices and put in nominal prices for the others.
To generate the unweighted price score for each service package the authority used the following formula (‘SP price bid’ represents the particular bidder’s price in the relevant Service Package):
Price = (1-2 × (SP price bid)/(Total of prices bid for that SP) )×100
Applying the calculation to the example prices gives the following unweighted price scores (rounded to the nearest whole number):
|
|
Service package 1 scores |
Service package 2 scores |
Service package 3 scores |
Service package 4 scores |
Bidder’s total scores |
|
Bidder 1 |
58 |
-2 |
52 |
11 |
119 |
|
Bidder 2 |
66 |
8 |
42 |
-7 |
109 |
|
Bidder 3 |
-24 |
95 |
6 |
96 |
172 |
In this example even though Bidder 3 has the most expensive overall score the decision to assess price solely at the level of service packages means that they obtain the highest score by a significant margin.
In the Proximus case, service packages were differently weighted and, once weightings were applied, each bidder’s score from each service package was added up to calculate the total price score for each bidder. As in the example above no global assessment of the overall price was undertaken. Overall Proximus’ price was the lowest bid. However once the weighted service package scores were totted up Proximus were not awarded the highest score for price.
Proximus’ case
Proximus brought proceedings alleging that the formulae used to calculate the price scores failed to identify the most economically advantageous tender and was unlawful as a result. There are some similarities to the Malmö case[1] in which the Swedish Court held that a formula which failed to identify the bid with the cheapest price overall as the highest scorer in the price criterion breached equal treatment and was unlawful. By contrast, in this case, the General Court ultimately had little time for such an argument but the judgment covers some interesting ground. Proximus put forward three arguments to challenge the price evaluation:
Argument 1: price should have been assessed globally, not solely at service package level
Proximus argued that because the opportunity was not split into Lots the financial evaluation should have been conducted globally rather than at the level of each service package.
The Court found that the authority’s choice to assess price at service package level was justified by the nature of the services and the objectives being pursued by the authority. The authority deliberately chose to split the specification into service packages and required bidders to price against each one so that they could ensure that the appropriate level of resource would be allocated to each part of the services. The Court found that meant that the evaluation criteria were tied back to the objectives being pursued under the procurement and dismissed Proximus’ claim on this ground.
In any event, this argument was an uphill struggle as it was clear from the procurement documents how financial evaluation would take place and Proximus raised no complaint until they lost.
Argument 2: the mathematical formula
Proximus argued that the only justifiable formula would be one that compared the price bid by the tenderer with the lowest price bid. This is a method commonly used to assess price. There are several forms of inverse proportionality formulae, which rank bids by their relationship to the lowest price. By contrast, the approach of the authority in this procurement was to evaluate bids by ranking them against the sum of all bids for that service package. Proximus argued that the approach followed here disproportionately rewarded bidders who bid low (relative to others) on the ‘financially less important service packages’ whilst failing to take into account overall price differences.
Proximus used several examples to demonstrate that the formula could have perverse results. In particular, as per the example above, they showed that if a bidder shifted the costs of a particular service package (or packages) into its bid for another service package (or packages) they could game the formula such that they could offer a higher overall price but still obtain a high price score. The Court gave this example short shrift – agreeing with the authority’s submission that it would have used its power to reject abnormally low tenders in that situation. Given the difficulties around the question of whether a tender is abnormally low that approach may raise further questions for contracting authorities in the future. It is an open question whether a bidder submitting a higher price than Proximus, but a £0 price in respect of one of the service packages might have had reasonable grounds to challenge had they been excluded for submitting an abnormally low bid and seen the contract awarded to a competitor whose bid was lower overall.
The Court had little sympathy for Proximus’ examples. This was so even where they demonstrated that including rogue bids could alter the spread of scores in an unfair manner. The Court was content that if the formula identified the lowest price bid for each service package as the highest scoring one it did its job ‘even if the scores awarded do not reflect the actual price differences between the tenders.’
Argument 3: the information not known to bidders
Proximus’ final attempt was to argue that crucial information was withheld from it, which meant that the formula was unfair. The crucial information was the sum of all bids for the particular service package. This could not be known by any bidder before submission. This is relatively well-trodden ground and the Court gave it short shrift: the criteria were clear in the tender specification, the formula was set out and explained in the procurement documents. All bidders were equally in the dark and as the sum is automatically calculated and applied irrespective of the result it had no discriminatory effect.
Overall
Although the choice of formula is well established to fall within a contracting authority’s discretion (subject to general principles) its application can easily have unintended consequences if careful modelling is not done beforehand.
In this instance, although the authority’s formula fell on the right side of the line it provided enough confusion that Proximus brought a sustained challenge, in court proceedings only finally determined over 2 years after the financial evaluation was completed.
Proximus might well have felt that, as the lowest priced bidder, they were entitled to be awarded the best price score. However, ultimately the Court was satisfied that the authority had sound reasons for the way it evaluated price. It is an interesting open question as to how the abnormally low tender approach the Court viewed as controlling for price manipulation in these circumstances would play out in reality.
The key lesson for contracting authorities is to give detailed consideration to the method of evaluating price to ensure that it delivers a robust and defensible outcome and to ensure that, as here, it is applied scrupulously fairly as between bidders.
