Towards a new method of calculating energy prices
CBS publishes monthly figures on inflation. Every month, inflation is measured as the rate of increase in the consumer price index (CPI) relative to the same month in the previous year. The CPI reflects the price development of a basket of goods and services as purchased on average by Dutch households. CBS therefore observes the prices of these products as they are offered at that particular time. Examples include food products, personal care products, hotel stays, clothing and also energy.
Current observation of energy prices
Inflation is currently at a historically high level. This is largely due to the sharp rise in energy prices. At present, that rise accounts for more than half of the current inflation rate. The prices of gas and electricity are observed monthly on the basis of all new (fixed and variable-rate) energy contracts as can be concluded at that particular time, similar to a number of other European countries. However, not everyone has a new contract at a new rate at the same time, and CBS did not have any data on which energy contracts were concluded at what time and at what rate.
This means the price information obtained through the current observation method is actually ahead of the average price development of energy for households in the Netherlands. The observation of only new contracts overestimates the average price increase when energy prices rise; it underestimates the price index when energy prices fall. The current method used for observing gas and electricity prices is adequate in a reasonably stable market, but in a volatile energy market with extreme price increases or declines, there is a need to refine the method.
Research into new observation method
To this end, in early 2022, CBS launched a study to find out in more detail at what prices households in the Netherlands buy electricity and gas. This study also takes into account already existing longer-term energy contracts, to better reflect the large discrepancies among prices paid by Dutch households, particularly between new and already existing contracts.
CBS has obtained a vast amount of data from a number of energy companies. Combined, these companies cover approximately 75 percent of all households in the Netherlands to which energy is supplied. CBS is still in talks with other utility companies in order to increase this coverage.
The data files have largely been analysed in detail and assessed for quality. At this time, however, the data are not yet suitable for use in the monthly CPI calculation. The new data are being used to observe prices of new and existing contracts. Based on this observation, a number of calculation methods have been compared, ranging from traditional methods with fixed annual weights to methods that take into account shifting between contracts and utility companies.
The final step in the study is implementation of the new method. Furthermore, CBS is working on a research report describing the data, methods, results, impact and considerations for the CPI.
Initial results of the methods examined
The initial results of the study show that, using observation methods based on new data, the price increases of gas and electricity as of autumn 2021 are significantly lower than under the current observation of energy prices. As a result, inflation would also be lower than what was published over the past few months. However, these analyses are based on data that are not yet fit for statistical production and are merely intended to provide an initial picture of what energy price developments and inflation might have looked like using the newly obtained data from utility companies.
The charts below show the published price developments of gas and electricity and the published inflation rate along with the lower and upper bounds of the range according to the newly examined methods. This range may vary from month to month. The charts represent the study to date and the results run from January 2021 to August 2022 inclusive. There is still insufficient data to make a conclusive statement about the months thereafter.
| Published (y-o-y % change) | Range of results under the new methods (y-o-y % change) | ||
|---|---|---|---|
| 2021 | January | -2.3 | -2.3 - -2.3 |
| 2021 | February | -0.8 | -2.0 - -1.3 |
| 2021 | March | 1.1 | -1.0 - 0.1 |
| 2021 | April | 2.8 | 0.1 - 1.4 |
| 2021 | May | 3.3 | 0.8 - 2.0 |
| 2021 | June | 5.1 | 1.3 - 3.0 |
| 2021 | July | 12.0 | 5.1 - 8.4 |
| 2021 | August | 13.4 | 5.4 - 9.1 |
| 2021 | September | 18.3 | 5.6 - 12.3 |
| 2021 | October | 30.6 | 6.0 - 20.9 |
| 2021 | November | 53.0 | 6.7 - 33.7 |
| 2021 | December | 66.5 | 9.2 - 46.4 |
| 2022 | January | 86.3 | 5.9 - 45.1 |
| 2022 | February | 76.2 | 11.0 - 41.7 |
| 2022 | March | 161.4 | 14.5 - 83.6 |
| 2022 | April | 140.8 | 21.7 - 76.8 |
| 2022 | May | 109.5 | 25.7 - 64.2 |
| 2022 | June | 82.7 | 27.9 - 53.6 |
| 2022 | July | 114.2 | 31.2 - 67.0 |
| 2022 | August | 170.1 | 34.0 - 96.5 |
| 2022 | September | ||
| 2022 | October | ||
| 2022 | November | ||
| 2022 | December |
| Published (y-o-y % change) | Range of results under the new methods (y-o-y % change) | ||
|---|---|---|---|
| 2021 | January | -9.1 | -9.1 - -9.1 |
| 2021 | February | -5.2 | -8.5 - -6.8 |
| 2021 | March | -2.4 | -7.2 - -4.6 |
| 2021 | April | 0.8 | -5.2 - -1.9 |
| 2021 | May | 3.0 | -3.7 - -0.1 |
| 2021 | June | 6.3 | -2.5 - 1.9 |
| 2021 | July | 15.8 | 2.5 - 8.1 |
| 2021 | August | 16.3 | 2.7 - 8.1 |
| 2021 | September | 24.4 | 3.1 - 13.6 |
| 2021 | October | 39.6 | 4.2 - 23.9 |
| 2021 | November | 74.9 | 5.7 - 44.8 |
| 2021 | December | 103.8 | 10.6 - 68.2 |
| 2022 | January | 110.7 | 8.6 - 48.4 |
| 2022 | February | 94.4 | 15.4 - 44.3 |
| 2022 | March | 181.9 | 18.8 - 88.7 |
| 2022 | April | 155.8 | 25.1 - 81.7 |
| 2022 | May | 118.0 | 28.9 - 68.1 |
| 2022 | June | 99.7 | 30.7 - 63.7 |
| 2022 | July | 116.6 | 21.1 - 71.7 |
| 2022 | August | 149.6 | 24.4 - 92.8 |
| 2022 | September | ||
| 2022 | October | ||
| 2022 | November | ||
| 2022 | December |
| Published (y-o-y % change) | Range of results under the new methods (y-o-y % change) | ||
|---|---|---|---|
| 2021 | January | 1.6 | 1.6 - 1.6 |
| 2021 | February | 1.8 | 1.7 - 1.8 |
| 2021 | March | 1.9 | 1.8 - 1.8 |
| 2021 | April | 1.9 | 1.7 - 1.8 |
| 2021 | May | 2.1 | 2 - 2.1 |
| 2021 | June | 2 | 1.8 - 1.9 |
| 2021 | July | 1.4 | 1.1 - 1.3 |
| 2021 | August | 2.4 | 2.1 - 2.2 |
| 2021 | September | 2.7 | 2.2 - 2.5 |
| 2021 | October | 3.4 | 2.5 - 3 |
| 2021 | November | 5.2 | 3.4 - 4.4 |
| 2021 | December | 5.7 | 3.4 - 4.9 |
| 2022 | January | 6.4 | 3.5 - 4.8 |
| 2022 | February | 6.2 | 3.9 - 4.7 |
| 2022 | March | 9.7 | 4.7 - 7 |
| 2022 | April | 9.6 | 5.6 - 7.3 |
| 2022 | May | 8.8 | 6.1 - 7.2 |
| 2022 | June | 8.6 | 6.7 - 7.4 |
| 2022 | July | 10.3 | 7.4 - 8.6 |
| 2022 | August | 12 | 7.5 - 9.6 |
| 2022 | September | ||
| 2022 | October | ||
| 2022 | November | ||
| 2022 | December |
The published figures and the figures as they follow from the study with new sources differ significantly. There are several reasons for this. One reason already mentioned is the fact that in the current method CBS only includes the prices of new contracts. In addition, price increases in variable contracts at energy companies are not always passed on directly to their customers.
The margins of the results with the new methods in gas and electricity price developments are wide. This is because the data and analysis still contain a number of uncertainties, such as:
• Incomplete observation. Current transaction data cover about 75 percent of the market.
• Selectivity of observation. Weighting methods have been applied to correct for the fact that 75 percent of the market is observed. While this coverage of contracts can be considered quite high, it does not guarantee correct price trends. The remaining 25 percent may have a different distribution in energy contracts and a different price trend.
• Uniformity of data delivery. The companies provide basically the same data, but since they are based on different ways of customer record keeping, there may be differences, which are not yet known.
• Incomplete data. CBS receives a cut-off from customer records at a certain time of the month. New contracts concluded after the cut-off point are missing from the data.
• Index methodology. Several methods exist to calculate a price index series using transaction data. Which method is most suitable depends on the data that eventually become available and also on how, for example, energy price caps feed through into average energy tariffs. Due to uncertainty about the future development of both aspects, different price index methods have been applied in this study.
CBS is currently investigating how to resolve these uncertainties and which method is most suitable for implementation. Developments around energy and available data are still evolving. For instance, it is still unclear how price caps can be properly incorporated under a new method and whether the new data sources will still be sufficiently suitable for use in the new method(s) studied.
Implementation of a new method
For the purpose of selecting the most suitable method, a number of criteria have been defined, such as practicality, robustness over time and alignment with EU regulations. The introduction of a new and more sophisticated methodology for observation of energy prices is likely to take effect in mid-2023, both for the CPI and the HICP. Inflation rates already published will not be adjusted at the time of the switch to a new method.
The selection of an appropriate method for measuring energy price trends is a challenge at the European level as well. The energy markets in the various European countries are very different and difficult to compare. Still, a number of countries have an energy price observation method that is similar to the Netherlands.
CBS is consulting with other countries in Europe and with Eurostat, the European statistical office, on the most preferred method of measuring energy price fluctuations. This is important in order to enable a proper comparison of the figures according to the European harmonised index of consumer prices (HICP). CBS would like to see more harmonised methods as soon as possible.
Regular progress updates
In the coming period, CBS will be providing regular updates on its research progress along with the news releases on inflation. For instance, at the end of November, when reporting the first estimate of November inflation according to the HICP, more information will be provided about the incorporation of energy subsidies and price caps into the inflation figures. Incidentally, such energy measures will not only be included in the new methodology but, where applicable due to prior implementation, also be incorporated in the existing methodology.
Research on the new observation methods is expected to be well advanced by January 2023. At that time, more clarity can also be provided on the observation method of energy prices that will be implemented and the date at which the new method is being implemented.
It has to be kept in mind that the results from research using a new method or multiple methods will only become available several months after the reporting month.
Flash estimates of inflation according to the CPI
CBS has been publishing flash estimates of inflation according to the HICP for some time. Flash estimates of inflation according to the CPI will also be introduced in early 2023. These will be published at the same time as the HICP and will concern December figures. Up until now, inflation according to the CPI is being published several working days after the HICP flash estimate. Flash estimates contain provisional figures only.