Energy is central to nearly all aspects of everyday life and business activity. Therefore, developing a long-term energy plan is most successfully achieved in a collaborative way that involves all government ministries, as well as outreach to the public and businesses. Producers of energy statistics are a key element in this process.

There are two key aspects of this collaboration: first, statisticians will provide the essential information needed to understand the current energy situation. These data need to be used effectively in the policy planning process and in the ensuing modelling work. Thus, it is vital that the statisticians are effectively integrated in policy planning and later in the policy monitoring processes that are ultimately established.

However, there also needs to be collaboration within the statistics community. In general, government statistics are organised in two ways: centralised or decentralised. For centralised information, the National Statistics Institute (NSI) is responsible for collecting and disseminating information to users, including ministries. Decentralised statistics will see ministries collecting and compiling data themselves, although ideally in co‑ordination with the NSI. There are pros and cons to both approaches. A more centralised approach allows for greater resilience and the sharing of knowledge across a wider statistics workforce. A decentralised approach brings policy colleagues closer to the data, but it can also mean that statisticians in ministries may be less aware of wider statistics developments or less able to maximise the benefits of wider statistical surveys.

A large statistical institute will still need a strong push to communicate across all topic areas, and will still need to reach out to the respective topic policy makers to ensure the statistics accurately reflect the situation in the country. Where a system is less centralised, statisticians will need to communicate and co‑operate across ministries to achieve a cost-effective approach to working, sharing ideas and informing colleagues of policy changes.

Fundamentally, structure is less important than achieving good co‑operation among statisticians, and effectively working together and enhancing links with policy colleagues (something that should be pushed from both sides). An advisory group involving all the relevant stakeholders has proven to be an effective channel for the countries to co‑ordinate activities around energy statistics.

A further beneficial condition for ensuring comprehensive energy statistics is the degree to which they are produced independently. There are several ways of achieving this, either via a Statistics Act or via codes that govern the way statistics are produced and disseminated. To whom the NSI reports in a government structure can also be important, especially in a centralised approach. In general, the closer the NSI is to the heart of government, such as a president or prime minister, the more likely it is that statistics are developed to meet the needs of a whole country across all topics – especially if independently governed.

One essential part of statisticians’ responsibilities is to measure and assess what happens as a result of the policy, via a monitoring process. However, statisticians and statistics should be involved across the entire process for if they play only this one role, then the policy design will be less comprehensive with less optimal outcomes.

The policy delivery cycle, as shown in the figure above, has a role for statistics and statisticians in all stages. Starting from the initial understanding of the situation, statistics are vital to show what is actually happening, and to explain why there may be unknowns, via missing data. This will help policy makers understand how the impact of a policy could be measured. In this first stage, as well as in the second stage when policy makers develop and appraise options, it is helpful and often necessary to review what other countries have done. Energy data are also a boon here, as they allow both statisticians and policy makers to understand the impact of similar policies.

It is always important to start with a defined benchmark, or the base data year from which change can be measured. These data may already be collected in energy statistics, but it may be necessary to produce a new data series. That work needs to start very early in the policy cycle, and to be complete by the time policy makers are in stage three, preparing for delivery. During this phase statisticians should be finalising the means of collecting the data to monitor the policy. This is likely to involve discussions with policy advisers as well as implementing agents, in order to have the correct data recorded that will make up the key administrative information captured as part of the policy. Should a policy be piloted, then statisticians should be involved in assessing the results of the pilot. Finally, as the policy is launched, then statisticians should continue to play a key role, ensuring the effective monitoring of the policy so that its impact can be properly understood.

Energy modelling remains an important part of the policy development process. There are virtually as many energy models out there as there are policy questions to understand. The choice of which model to use is driven by the needs of users and depends on the questions to be answered. Often a single model is not enough to address all relevant aspects of a question or scenario.

Models can be used to provide projections and explore scenarios. They can answer questions such as what will happen in 2050 with today’s existing policies. They can also explore normative scenarios such as how to reach a single (or a set of) policy target(s), or explorative scenarios such as understanding the impact of specific policy measures or strategies on the energy sector. Often these are compared against a baseline or business-as-usual scenario. Some models can be very complex and only usable by trained users, but some are designed to be used by a wider population, in order to open up debate in government, media, academia and the investor community. These models do often require a lot of work outside of the model itself. Energy models can be top-down or bottom-up. They can aim to model everything in the energy system and find cost-based solutions or they can be used to explore more speculative scenarios.

In some cases models can be used to estimate missing data on a more granular level than provided in energy balances, or check the plausibility or consistency of data from different sources. In general, though, direct collection of such data is preferable.

As a generalisation, models in wide use can be thought of as being one of four types: general equilibrium (GE) type models; MARKAL/TIMES; LEAP and 2050-type calculators. The figure below highlights some of the differences between these models based on three factors.

For a more detailed description of model types, please see the annex.

Models can also be distinguished along the dimensions of simulation versus optimisation and energy sector versus economy-wide models, designed around the questions that different fundamental modelling approaches can address.

GE type models cover the entire economy, while energy system models such as LEAP and TIMES cover only the energy sector (for that reason, the latter are also referred to as “partial equilibrium models”). While the energy sector is not represented in a technologically detailed fashion as in energy system models, GE models help to analyse the interactions of the energy sector with the rest of the economy, e.g. the impact of energy policies on GDP, employment or trade balances.

Some models are used widely around the world, for example LEAP and calculator type models, given that these are available for free in developing countries. However, the software costs alone do not reflect the full costs of energy modelling – the most expensive aspect is the staff needed to do the modelling. Staff fluctuation, particularly in government ministries, has often hampered the development of long-term in-house modelling capacity in developing countries.

What is key to remember is that no one model type is better than another. Finding the right model is driven by the needs of the country. And since not all models can do all things, combinations of models can be very beneficial. For example, a country may want to use a 2050-type calculator to have a wider debate about possible policy choices, but then may need to make use of a MARKAL/TIMES model to properly understand the costs of the chosen route. An approach like this may well be a good route for Azerbaijan.