Applied Game Theory – the future of IT procurement?

August 17, 2020

DLA Piper’s Technology and Strategic Sourcing team recently completed a circa £275 million transformative networks and telephony deal for one of its key clients (one of the UK’s largest financial institutions), who elected to utilise an innovative negotiation method known as ‘applied game theory’ in order to carry out the final negotiation stage of the procurement process. Game theory is a Nobel Prize-winning branch of behavioural economics that was pioneered by Hungarian-born American mathematician John von Neumann in 1944. It is centred on the scientific modelling of interactions between multiple self-interested and rational parties, who engage with each other based on how they think the other parties may act. More familiar examples of game theory in action, or applied game theory, are chess or the “prisoners’ dilemma”, but applied game theory has more recently emerged as a tool to predict behaviour and steer negotiations in the commercial world. In this paper, we compare this applied game theory approach to more traditional negotiation methodologies, seeking to reflect on its advantages, disadvantages and common pitfalls, and set out some tips for anyone else looking to use this method of procurement.

Applied Game Theory vs ‘traditional’ negotiation methodologies 

Any contract negotiation methodology should aim to: (i) deliver a ‘good’ agreement which is robust but not excessively burdensome for either party, (ii) be efficient in terms of time taken and effort required, and (iii) improve or at the very least maintain existing vendor-customer relationships. A traditional, large-scale multi-bidder negotiation would usually involve the issuing of a request for proposal (“RFP”) pack, following which bidders would respond with a mark-up of the contract together with any other relevant documents. An initial comparison of bidders’ positions would then be undertaken by the customer. Depending upon the customer’s specific procurement requirements (cost, time, resource for example), parallel negotiations between the customer and the multiple bidders would then normally ensue, with down-selection(s) made as the negotiations progress based on a set of pre-defined evaluation and scoring criteria. Remaining bidders might then be required to submit a best and final offer (“BAFO”) before a final contract award is made. In preparing for a negotiation, we often advise our clients to ensure that they also have a minimum viable position (“MVP”) that would be acceptable as a basis to move to concluded contracts, and a best alternative to a negotiated agreement (“BATNA”) in the event that an agreement cannot be reached via the negotiation.   

Game theory can fit within such an approach but in effect replaces the BAFO stage and instead involves the scientific modelling of the interactions between different self-interested parties (the ‘players’), and sets out clear ‘rules’ for a game (in this context, a negotiation) in which the players ‘play’ against each other, with the ‘winner’ then ending up being awarded the relevant contract. The clearly defined rules and processes create a forum in which the behaviours of the different players can steer others and be steered towards a more desirable outcome, adding more structure and transparency, and reducing the risk to a procurement process. 

Game theory in practice 

We initially adhered to the more ‘traditional’ negotiation process detailed above, up until the point at which there were no remaining ‘deal breakers’ but a number of material issues remained. At this point, ‘traditional’ negotiations and documents were ‘frozen’ as preparations were made to undertake o the applied game theory process. 

All parties were briefed as to how the process would be run and were required to formally accept the pre-defined process (devised by the third-party consultants) and its outcome. Each of the client’s project workstreams (Commercial, Technical, Finance, Transition & Transformation and Legal) were required to categorise all outstanding issues with the remaining three bidders into ‘material’, ‘nice-to have’, and ‘non-material’ buckets. Those issues deemed ‘material’ were allocated a financial penalty (a positive GBP value) determined by the relevant workstream and moderated across the piece. Conversely, any particularly impressive or favourable elements of a bidder’s proposal (such as  a particularly innovative solution) were awarded bonuses with a negative financial value (in GBP). These bonuses and penalties were then combined and applied to each bidder’s financial offer/pricing, so as to then produce an adjusted ‘comparison price’, which was said to represent the ‘true’ value of the contract, having taken into account any key issues or bonuses in the bidder’s overall proposal as well as the actual contract price. 

The bonuses/penalties and comparison price were then revealed to each of the bidders (i.e. each bidder saw their own bonuses and penalties, but not those applied to the other bidders), who were subsequently given a short time period in which to improve their comparison price by withdrawing (or improving) issues giving rise to a penalty and requoting their solution accordingly. At this stage there was already clear potential ‘winner’, based on who had the lowest comparison price. 

All parties were invited to a ‘final negotiation day’, which made up the last stage of the applied game theory process. This was a time-boxed, face-to-face set of meetings, during which each bidder was approached by the client, starting with the bidder with the lowest comparison price. This bidder was then given the opportunity to try and close the deal first, by accepting a Take It Now Offer (“TINO”) comprised of a number of side conditions specified by the client, which were based on any remaining unresolved material issues. In the event that the leading bidder did not accept the TINO and thus close the deal in a set time period allotted to them on the day, the bidder in second place (based on their comparison price) was approached and the client again attempted to reach agreement. If the deal could still not be closed, all parties would move to a sealed bid offer process. In our case, and after some further negotiation around the specifics of the TINO, a final deal was struck later that evening, without having to move to the sealed bid process.

the game theory negotation process

Fig 1: a time line of the process

Reflections

All parties recognised that the applied game theory process was by no means a perfect one, and as with every new process, had room for improvement. This was also the first use of the applied game theory on such a large scale by our client. Despite this, it was viewed by the executive stakeholders as an overall success which made the contract-award decision-making process more transparent and objective for both the client and bidders and resulted in a better deal overall for the client.

Attributing a monetary value to the outstanding material issues was undoubtedly one of the more challenging aspects of the process but was of paramount importance as it directly impacted a bidder’s final comparison price (and therefore its overall ranking). The relative ease of producing such values depended heavily on the workstream.  This was more straightforward for a Commercial issue. For example, here  the bidder did not agree to providing an annual innovation fund of £10,000 and therefore received  a penalty of +£10,000 which was added to their comparison price: this penalty would reduce by £2,000 for every £2,000 the bidder agreed to put towards an innovation fund. Conversely, for the Legal workstream, producing a monetary value for a material issue linked to the limitation of liability cap, for example, was more difficult. It is of course not possible to predict the probability of such a clause being relied upon in the future (and therefore the likely impact of having a higher or lower liability cap). Assigning a value to such a risk is difficult, short of having a crystal ball or a reliable set of historic data that could predict the likelihood of this occurring in the future based on the past. In order to attribute an appropriate value, we introduced penalties for positions that deviated from standard market positions for similar transactions and ensured that these penalties were not disproportionate to the penalties allocated by other workstreams. 

When bidders were later presented with their penalties and bonuses, some challenged the quantum and requested further information as to how it had been calculated and how they could reduce it. Another bidder thought that their bonus was undervalued and wanted a further chance to demonstrate their capabilities in this regard. The process did not allow sufficient time for either of these opportunities so we would recommend that any future use of the applied game theory process allows for such a mediation or discussion between parties around the bonus/penalties. This could reveal helpful clarifications and ultimately result in a more accurate comparison score. 

The applied game theory methodology is undoubtedly (and has been proven to be) an effective approach for commoditised tenders, where the product across bidders is broadly uniform and the bids are largely evaluated on price. However, we were applying the theory in the context of a large, complex and multi-faceted transaction, where proposals were evaluated in numerous ways: there were many moving parts and solution nuances, which therefore presented some challenges. For example, despite our best efforts, it was not feasible to have the ‘final negotiation day’ as the truly final discussion on a number of the outstanding issues, and owing to the nature of the transaction, we were still resolving issues and contending with new issues that arose right up until the date of signature. While  the game theory process made use of  fully quantitative approach in order to evaluate which bidder should win, a more traditional negotiation process would ordinarily and inherently have included some qualitative analysis as well (such as the cultural ‘fit’ of the bidder with the customer), which this process lacked. We would recommend that qualitative measures are factored into use of applied game theory where relevant by attributing bonuses/penalties for such qualitative aspects, in the same way as the quantitative ones. 

Tips for use 

For those looking to utilise applied game theory in their future procurement processes, we would recommend the following:

1. One size does not fit all 

Consider carefully whether your transaction is appropriate for the use of applied game theory and whether the nature of such a process aligns to your interests as a customer. For example, if the underlying incentive behind the transaction is solely cost saving, this may be a useful methodology for you. However if the transaction is particularly complex, it may be best to use this only at the very end of the procurement, if at all, when a small number of key issues remain and with an acknowledgment that further discussion and even negotiation may be required following the use of the game theory process. 

2. Communication is key 

Remember that you will be entering into a long-term arrangement with one of your bidders so ensure that you build good foundations for this relationship during the procurement process. All stakeholders involved in utilising applied game theory must understand how the process will work, the purpose behind using it and how it will feed into the wider procurement process. 

3. Keep it competitive

While trying to ensure transparency to bidders during the process is important to help incentivise them, this must be balanced with retaining as much competitive tension as possible. Don’t show your hand by revealing internal pressures (such as the use of timeboxed negotiations and down-selection dates) too early, or even reveal the finer detail as to how the game theory scoring/price adjustments will work, as you may otherwise find that the bidders will actually stop negotiating and instead ‘store up’ issues for trade off on the final day. Ensure you maintain at least a two horse race until most, if not all, material issues have been agreed and allocate costs and resources appropriately in order to do so. Equally, bidders are unlikely to fully engage unless they truly believe they have a chance of winning, so take care not to cross the line and push a bidder so hard that they walk away – awarding measured bonuses as well as penalties will help to minimise the risk of this happening.

4. Qualitative vs Quantitative analysis 

If you opt to use the process, ensure that when planning to do so you consider exactly what your main objectives are for the procurement: an innovative solution: keeping costs down or the lowest risk option? These should form both the qualitative and quantitative evaluation criteria you use when evaluating the bidders, as once  parties agree to be bound by the decision of the process, you will be bound by these criteria to determine the contract award. In the event that these criteria are not considered carefully, the result could be that a party could effectively ‘game’ the applied game theory. They could make a particular penalty especially punitive, or not allocate a fair bonus value to a particularly good aspect of a bidder’s proposal, so as to counter for those qualitative aspects that should otherwise have been included. 

5. Prepare to be flexible

A common failure in negotiations is a failure to be flexible and this is just as relevant in relation to applied game theory. In our transaction, while all parties agreed to use the game theory process and to be bound by the decision resulting from it, bidders wanted more time to discuss with our client the reasoning behind certain penalties and to demonstrate their capabilities in other areas where they felt they had not received enough recognition. Despite project