Modelling the complex path to net zero

Modelling the complex path to net zero

A bottom-up model of consumer behaviour yields rich results

Fortunately or unfortunately, depending on your taste, people have a habit of not behaving in a rational manner that economists expect of them. This refusal to conform is a headache for governments as they plot strategies to wean society off fossil fuels.

Models of the transition to net zero carbon emissions in the heating sector by 2050 typically focus on the socially optimal pathway – i.e. the technologies that should be adopted to minimise overall costs while meeting various constraints, such as ensuring that demand can be met while respecting carbon targets.

The problem with such top-down models is that consumers do not always act as the economic textbooks predict. So unless the government were to mandate the specific technologies to replace oil- and gas-fuelled boilers, the outcome will depend on the choices made by millions of individual households.

Consumers will not necessarily go for the lowest-cost option. Their choices may be affected by a host of factors, such as the hassle of switching to a new technology, or the extent to which they are familiar with a particular alternative. Some people will be influenced by what others are doing and will adapt their behaviour accordingly – a phenomenon known as social proof.

Deeply felt views about green technologies, for or against, might be another determining factor, as will differences in how people assess the upfront cost of a new low-carbon system versus the long-term savings it will generate. Consumers’ choices may also be heavily influenced by recommendations from heating system installers.

Modelling complex decision-making

A bottom-up, so-called “agent-based model” (ABM) differs from conventional top-down approaches by focusing on these individual decisions and, importantly, how different groups of agents interact with each other. The individual rules they follow are often very simple but can result in unexpected, complex behaviour.

With this in mind, the National Infrastructure Commission commissioned Frontier Economics to build such a model to see whether ABMs can yield insights into how the UK heating market may develop. While the project was not intended to assess specific policies, it can show how this type of modelling may be a useful tool for policymakers.

The decision-making process of property owners is the core of the model, which also incorporates the actions of other agents such as property occupiers,  installers and gas distribution networks. The decisions taken by these groups influence one another. By simulating these interactions, the ABM can illustrate the high-level outcomes that may emerge as a result. For instance,  the take-up of low-carbon heating technologies by some owners may spur others to follow suit.

Assessing the barriers to low-carbon heating

The insights provided by the ABM are apparent in a scenario which assumes that the only green source of heating comes from electrification, by deploying heat pumps (we also considered other scenarios where hydrogen boilers are an option). Under this scenario, gas boilers remain the dominant technology by 2050, despite rising carbon prices leading to increased running costs. Why? Because of the way households are assumed to be making decisions. When the model is run with the same cost assumptions, but with customers making up their minds purely on grounds of the lowest cost, the take-up of heat pumps is much higher. The model can therefore quantify the impact of non-monetary barriers to take-up and the effects if they can be overcome

Admittedly, the relative importance of different obstacles to low-carbon heating could be demonstrated without constructing a full ABM. The real value of such a model comes from being able to simulate the results when the actions of one group can affect another. This can lead to outcomes which could not be forecast when agents are considered individually.

The impact of these interactions can be substantial, our modelling showed. In one scenario, we simulated an intervention such as subsidies or a demonstration programme targeted at a subset of consumers. We found that, for every two customers directly influenced to take up a heat pump, an additional customer might take up a  heat pump as a result of indirect effects. These effects include the influence of customers on their neighbours, as well as the development of a local market for installers.

This sort of spillover can make policies more effective and is something that could not be captured by a conventional model that looked at the targeted set of property owners in isolation. Our work also suggested that focusing on a concentrated group of owners, say in a single town, might dampen the spillover effect. Spreading a major policy intervention over a wider area might achieve more bang for the buck.

An integrated, practical framework

Our project has demonstrated that agent-based modelling can provide valuable pointers to help with the transition to net zero.

This type of modelling is particularly helpful where the decisions made by customers or firms may interact with one another. As shown above, this can lead to situations where the cumulative effect of a policy is greater than the sum of its parts. ABMs can help best target policies to make the most of these feedback effects.

The very structure of an ABM, where the behaviour of each agent is made explicit, also provides a clear and objective framework around which to build the evidence base for consumer decision-making.