Although scientists have spent a considerable amount of time making long-term forecasts more reliable, they are little used by business. Close collaboration with the electricity sector has given us an opportunity to understand why, writes CICERO researcher and economist Asbjørn Aaheim.
Climate change was one of the main topics at this year’s World Economic Forum, where the world’s most powerful met to discuss what is happening in the world and how to face the future challenges. This illustrates an increased awareness within the world of business of what the consequences of climate change may be, and this has led to higher demand for information from climate science.
Against this backdrop, both scientists and those financing research have recognised that there is a possibility that knowledge about the climate system – which again forms the basis for understanding what could happen – could contribute directly to value creation in society.
For that reason, there have been several initiatives intended to encourage climate scientists to find industry partners they can cooperate with. Such cooperation will make businesspeople better informed and updated on what climate change could mean for them and enables scientists to get signals from business about what to focus on in order to make their knowledge more relevant for them.
Considered okay background information
The idea is good, and it would be strange if there was no potential in it. Nevertheless, in practice, it is difficult to find concrete examples of businesspeople using scientists’ estimates of future climate for making decisions.
The estimates are considered okay background information which can provide signals of the changes to come, but it is impossible to say anything about how data from these estimates impact business decisions.
One explanation is that estimates of future climate are usually made with a time horizon of 50 or 100 years, and it is difficult to say anything about the changes to come over the next 10-15 years, which is also far into the future for most companies.
For that reason, there has been a lot of focus on so-called seasonal forecasts, where models corresponding to the climate models are used to make weather forecasts for the coming weeks and months, and sometimes even for up to two years ahead.
Many decisions are weather-dependent
Many companies make plans within such a time horizon, and their decisions often depend on expectations about the weather to come. Hydropower producers, for example, must balance the water inflows to their reservoirs with the demand for electricity, which is also dependent on the weather – first and foremost on the temperature.
Farmers also plan their crops in light of their expectations for the weather over the next few months, and producers of typical seasonal products such as skis, ice cream or beer also need to take weather forecasts into account when they make plans for how to meet demand.
People in these industries would obviously appreciate getting information about the weather to come in advance. Nevertheless, their interest in seasonal forecasts has proved to be rather low, and their explanation for this is often that the forecasts are not reliable enough to be trusted, and that they trust their own evaluations more.
Many also point out that the weather is just one of many factors with which there is uncertainty, and it is difficult to say how they are combining quantified projections from the seasonal forecasts with qualitative assessments of other uncertain conditions.
Those who argue that seasonal forecasts should be used more actively seem to interpret this as a sign that businesses do not fully understand how they can benefit from using the forecasts. To show what their potential is, they study specific decision-making processes and calculate how much the companies could earn, or save, from making better use of the forecasts.
Many companies remain sceptical about seasonal forecasts
Such calculations have, however, not made companies less sceptical about using seasonal forecasts. No one disagrees with the fact that more reliable forecasts would have increased their utility, but those who make the forecasts still argue that seasonal forecasts are more valuable than what their current use would indicate.
Another explanation for the low interest may be that the information provided in the forecasts is not relevant to the decisions that companies make. Although the purpose of the close cooperation between the producers of seasonal forecasts and the expected users is to get increased clarity on this, it has been difficult to pinpoint which data users find that they can use actively.
Companies often make their decision based on their own experience and “gut feeling”, without having concerns about how and to what extent they are relying on quantified estimates when drawing their own conclusions.
Companies must decide when to act
An alternative way of defining what the information needs of the companies are, is to take as a starting point what information is needed to explain the decisions they make. You do not get this information from the seasonal forecasts the way they are currently presented.
In the seasonal forecasts, you get information on possible deviations from the norm for selected weather indicators, such as temperature and precipitation, over the coming weeks and months.
This is interesting information, for example if you are planning your summer holiday and are wondering whether to enjoy the summer at home or go somewhere where it is warmer and sunnier. Before you decide, however, you would often want to know more. A lot of things may happen before summer sets in, so it may be best to wait a while and see.
Hydropower producers make similar assessments every day: should they produce electricity now, at the current prices, or would it be better to save the water in the reservoirs for later, when power prices may be higher?
When the future is uncertain, having information that helps you decide what to do is not always the most important. Sometimes it may be more important to have information that helps you determine when to decide.
Organise the information for decision-making purposes
The way they are currently presented, seasonal forecasts can do little to help decision-makers determine when to make a decision. To do that, they would need a basis for weighing today’s forecasts against any potential updates over the coming weeks and months.
Some decisions make decision-makers commit to a certain choice for a long period of time. Wind power producers, for example, need to decide when to perform maintenance work on their wind turbines. Once maintenance work has started, wind power production at the turbine will stop for several weeks, maybe even months.
Other decisions give you the opportunity to change your mind if the weather is not as expected. If the wind power producers continue to produce wind power, they can later decide when to start maintenance work on their turbines.
Such decisions cannot be made solely based on the information available today on how the weather will be over the next weeks and months. You also need to consider the likelihood that the weather outlook will change as the seasonal forecasts are updated.
One thing that is quite certain is that the forecasts for a certain period, such as “next summer”, will be less uncertain the closer in time it gets. This means it is important to get an estimate of how much less uncertain future forecasts for that specific time period can be expected to be.
A trial-and-error process
The EU-funded project S2S4E has developed an online weather service featuring seasonal forecasts, which shows the global weather outlook for up to three months ahead. This tool has been created in cooperation with the power sector in order to ensure that the information is presented in a way that is useful to them.
In the last remaining stage of this project – which will be finalised in November 2020 – we will use the signals from the user partners involved to derive which information from the seasonal forecasts they need to make their decisions.
Taking forecasts for several different time periods as a basis, we will try to identify a possible systematic in the updates. This can involve how the insecurity has changed as the forecasts have been updated as the week or month that the outlook is for has gotten closer, or how or possibly also if the insecurity changes in different seasons.
Our proposal for how information from the seasonal forecasts can be organised for decision-making purposes will be based on decision theory. We cannot take for granted that the users will find this information useful for their practical needs.
However, the purpose with this proposal will primarily be to take a first step towards increasing the usefulness of seasonal forecasts for the users by taking as a starting point the information needed to explain the decisions that have been made.