Strategic Bidding in Flexibility Markets

The current transformation in the energy system, with volatile renewable feed-in and the electrification of demand, creates increasing stress on the electricity grid. While conventional power plants are no longer available for congestion management due to phase-outs, the opportunity costs of small plants and loads are often unknown, so they are not covered in the conventional redispatch processes. As one possible solution, market-based flexibility mechanisms have been evaluated lately to tap the flexibility potential of small generating assets and load-side flexibility. In the unIT-e² research project, we are, for example, exploring ways of using the flexibility of electric vehicles and heat pumps for redispatch via commercial aggregators. However, market-based redispatch mechanisms have raised the concern to incentivize a form of strategic bidding behavior, referred to as “increase-decrease gaming” (inc-dec). The suppliers of flexibility thereby expect a price on the flexibility market that is favorable to the price on the spot market and thus adjust their spot market bids accordingly. Possible consequences discussed are an increase in grid congestions, rising costs for the grid operator for flexibility procurement and undesirable investment incentives.

Understanding the Problem with the Interactive Gaming Dashboard

To illustrate the issue for various congestion scenarios, considering both generation and load-side flexibilities, and as a basis for discussions within the project consortium, we at FfE have developed a theory of choice model from the perspective of actors. In this model, flexibility providers align their bids in a profit-maximizing manner in both the spot and redispatch markets. The model is integrated into an interactive dashboard that allows users to assess their own scenarios, market design decisions, and mitigation strategies in terms of their impact on Inc-Dec Gaming.

This allows, for example, to understand the risks associated with strategic bidding. Let’s consider a load (e.g., the EV-fleet from the figure) whose true willingness to pay is above the spot market price: if it is awarded on the spot market, this results in a welfare gain for the load (consumer surplus). Additionally, if the load is located in a grid area that frequently experiences feed-in related congestion due to a high penetration of renewable generators, it can anticipate such congestion and hope to obtain the energy more affordably through a redispatch measure than on the spot market. This creates an incentive for the load to intentionally lower its true willingness to pay on the spot market or not submit a bid at all. However, if the congestion does not materialize, the load may also be unable to source energy through the redispatch market, resulting in a welfare loss compared to truthful bidding on the spot market.

What can be done about it?

The model calculations have shown that strategic bidding in flexibility markets is particularly critical when there are only a few participating actors, there are high price spreads in the spot market, and pivotal providers are present – in other words, when individual providers can provoke congestion through their behavior. To mitigate this risk, several mitigation strategies are currently being discussed, the effectiveness of which we have examined using the model:

1. Upstream Bid Fixation: This mechanism aims to force providers to submit their bids on the redispatch market well in advance before they could predict a potential bottleneck. While this can help restrain the exercise of market power on the redispatch market, the actual Inc-Dec Gaming occurs through the bids on the spot market. Even if the redispatch bid has been fixed for a long time, providers can still strategically position themselves on the spot market, thereby exacerbating the bottleneck situation.
2. Stochastic Non-Attribution: In this approach, an attempt is made to artificially increase the risk for providers on the redispatch market by introducing a certain level of uncertainty in the attribution process. This means that providers risk not being awarded even if they have the most competitive bid. However, this approach automatically incurs higher procurement costs, and in our scenarios, the non-awarding rates would need to be at least 48% to have a deterrent effect.
3. Capacity-Based Redispatch: The only measure that has proven effective in our model is compensation for redispatch based on capacity reservation rather than the strike price. In a pure capacity auction held before the spot markets, conventional Inc-Dec Gaming is not expected because only the provision of capacity is compensated. Thus, whether the congestion actually manifests does not affect the provider’s profit. Providers could, at most, increase their capacity bid in anticipation of congestion, potentially exploiting a dominant market position. However, this could be mitigated through upstream bid fixation. Importantly, there should be no arbitrage opportunities to the spot market during the actual dispatch. Therefore, increased loads should pay the current intraday price in the redispatch case, while load reductions receive compensation at the same level.

The results of these investigations contribute to several ongoing research questions within unIT-e² and other research projects at FfE on the subject. These include the design of Redispatch 3.0 and the development of a coordination platform for flexibilities in low-voltage grids.