Collective bargaining and a dose of game theory can help you lower your energy bills

Too much, almost inevitably. Rui Vieira/PA

Energy prices are rising, and it hasn’t gone unnoticed that the profits of the handful of large energy supply companies are rising too. While it can be argued that there is no direct causal relationship between the two, there is clearly a case for consumers to have access to better tools that help them access the best tariffs and lower their bills.

In particular, those consumers with energy consumption patterns that are more predictable or desirable for the companies that supply them should be able to demand better prices than others without. The problem is that a single individual customer doesn’t have much in the way of negotiation power in the market. One possible answer is for them to band together into groups, and engage in collective bargaining to get a better deal, a concept called collective energy purchasing.

The trend of joining forces to negotiate better prices for their electricity started in continental Europe, particularly in Belgium and the Netherlands, but has gained considerable traction in the UK. The Department for Energy and Climate Change ran a £5m fund in 2012 to provide support to organise such group bargaining.

Schemes such as the Big Switch and the Big Deal provided thousands of consumers with a way to switch, saving up to 25-30% on their annual electricity bill. More recently, several start-ups firms and local initiatives have started offering similar schemes.

A better deal for all

The process of joining together is typically mediated by a third party, and a large part of it can be automated. For instance, given the adoption of smart meters, it is not hard to imagine a near future in which consumers can simply upload (or provide access to) information about their usage, and a web service can work out which is the optimal tariff and carry out the collective purchasing and switch on their behalf.

It’s a compelling idea, but there’s no guarantee an automatically selected tariff will always be the best choice for all the group’s customers. Buying as a group may provide an optimal result for the group, but this is an averaged result rather than one that applies with respect to each individual customer. Individual customers (or subgroups of customers) may be better off switching individually, or forming their own subgroup around another tariff.

This type of phenomena is called coalitional stability, and has been long studied in coalitional game theory, and more recently, in distributed artificial intelligence. Recent research has started using AI techniques to address the challenge of designing more efficient group-buying aggregators.

Predicting use and spreading risk

One central issue is modelling how predictable each customer is. Distribution companies must estimate how much electricity their consumers will use and buy long-term, forward contracts. Any shortfall of electricity has to be bought on the spot market or during balancing, typically at a higher price. The converse also holds, in that any electricity bought in excess has to be sold during balancing, usually at a loss. So having accurate data about consumption is important.

A “prediction of use” tariff, which asks customers for a prediction of their electricity use (or estimates this from their past consumption records) and charges them accordingly would better match their cost to the supplier based on how predictable their energy use is. Crucially, while each consumer may be unpredictable, grouping them together in a collective reduces their aggregate uncertainty, making their consumption more predictable.

In fact, a market could comprise a whole range of these prediction of use tariffs. Some of could be flat, like existing tariffs, where the utility company would carry the risk but charge higher prices. Others would encourage greater predictability from customers, lowering the risk to the supplier who in turn provides a much better price for electricity consumption within the predicted limits (and extra charges for use beyond the predicted amount). Different consumers with different requirements could be dynamically clustered, depending on how well they predict their consumption, with buying groups formed around particular tariffs.

The sort of coalitional game theory that can help design software and tools to provide the best tariffs can also divide the bill in the most fair way. On way is the concept of marginal payment, where a customer pays the difference between what the group pays including him or her, and what the group would pay if he or she were not a member.

It’s conceivable that better artificial intelligence techniques can help us provide incentives for people to form groups. A well known problem in electricity group buying is that people are reluctant to commit until a critical mass is reached. So marginal payments could depend not only on ease of predicting consumption but also on how early a member joined their collective purchasing group.

Used wisely, such collective schemes can raise consumers’ awareness of their energy usage, lowering overall energy consumption, leading to less carbon emissions, lower costs for supplier and consumer, and less wastage.