Within the group of products with a low import content (NT), wages represent a large share of total costs for companies. This is because they are mainly made up of service-producing companies, where the share of labour is higher than in the goods-producing sector, for example. An approximation for volatility can be deviations from historical averages in terms of standard deviations.

Figure 4 indicates that the volatility of wages is low compared to other business costs. Due to limited data availability, the figure includes spot prices for electricity and oil costs and does not take into account fixed contracts. However, the standard deviations of these components are so high that it can be assumed that the volatility including contracts is also relatively high. Given the stickiness of wage costs, there may be reason to assume that price growth in NT is also stickier than in T.^[8] See, for example, the May 2019 speech “The rise of services and the transmission of monetary policy” by Benoit Cœuré. The speech points out that wages account for around 40 per cent of input costs in the euro area services sector, compared with only 20 per cent in manufacturing. In the services sector, the figure shows that only 5.6 per cent of prices change each month, compared with 9.2 per cent for non-energy industrial goods and 28 per cent for non-processed food. If input costs in the NT component are stickier than in the T component, this may mean that inflation is also stickier. An increased share of NT in the CPIF over time may thus have contributed to the rate of price increase when inflation has risen remaining at a high level for longer than it would otherwise have done.

Figure 4. Figure 4. Volatility of business costs in Sweden in terms of standard deviations Number of standard deviations from the historical average

One way of measuring the degree of persistence in inflation is to estimate equations, where the current rate of price increase is explained by the rate of price increase in earlier periods. The sum of the coefficients for past rates of price increase provides a measure of the degree of persistence that can be compared across different sub-groups of the CPIF.^[9] See Jesper Johansson and Oskar Tysklind "Characteristics of subgroups in the CPIF", Economic Commentaries no. 9, 2024, Sveriges Riksbank. Figure 4 shows the sum of the estimated coefficients for the first six own lags for seasonally adjusted monthly changes.^[10] The seasonal adjustment method used is Census X-13 in Eviews. Own lags are monthly changes in the series. The purpose of these exercises is thus to get a sense of whether, on average over a long period of time, a price increase is followed by price declines or moderate increases, which in such cases indicates a low persistence. The opposite case is if the price increase is followed by continued price increases or a continued high price, which in such cases indicates a high persistence.^[11] For a review of the calculation procedure and the arguments behind it, see Jesper Hansson, Andreas Johnson and Sara Tägtström "How persistent is inflation in Sweden?", Economic Commentaries No. 5, 2009, Sveriges Riksbank.

The results in Figure 5 do not indicate that NT (“Rents” and “Other services”) has a higher persistence than most components of T (the other five groups). This is despite the persistence in particular of “Rents” when measured in this way appearing to have increased since 2010. The persistence of prices in the “Electricity” group is particularly high when measured in this way. The coefficients sum to close to 1, which according to this methodology indicates a high degree of persistence. An interesting aspect is that the persistence measured in this way increased for all components except for “Electricity” where it decreased marginally. However, it is difficult to explain why this clear broad change has occurred. One potential explanation could be that an increase in the share of services in the economy with a higher degree of wages in the cost base has increased the persistence. However, measured in this way, it indicates that "Rents" and "Other services" have, on the contrary, a relatively low persistence.

Figure 5. Figure 5. Estimated coefficients for own lags for different components of the CPIF Sum of the first six estimated coefficients for own lags

The results of these exercises thus indicate that the NT components are not very persistent and that, for example, electricity and gas prices are. These exercises take into account all monthly changes over the entire periods in question and not for specific inflation cycles. The persistence of electricity and gas prices may be somewhat surprising given the large fluctuations in energy-related prices in the CPIF in recent years. Another method of analysis is to look at the main upward and subsequent downward movements in inflation that have occurred and analyse which components fall first and which lag behind. The criterion for a rise in inflation here is that it peaks at just over 3 per cent, i.e. clearly above the Riksbank's inflation target of 2 per cent. Figure 6 shows the development of inflation in the CPIF, the product category with a low import content and the category with a high import content as an average of two inflation increases and declines during the time the Riksbank has had an inflation target. During these two inflation upturns and declines, CPIF inflation in terms of monthly data, in February 2003 and September 2008, peaked at around 3.5 per cent. These calculations can be compared with the latest cycle shown in Figure 7, where data are shown from the first quarter of 2021 up to and including the latest outcome, the first quarter of 2024.

There are three observations that can be made in Figures 6 and 7. The first is that the rise in inflation was much faster than normal in the latest cycle. It can also be observed that the rate of price increase in the product category with a high import content around 2003 and 2008 had relatively large and distinct upturns and declines on these occasions. At the same time, the rate of price increase in the low import content category did not change much, although the patterns are relatively similar for these cycles compared with the latest one. Four quarters after the peak observed in these two previous cycles, the rate of price increase in the category with low import content remained at a significantly higher level than for the category with high import content. The decline in the observed inflation for low import content is marginal and has occurred after cautious increases, indicating stickiness. In the latest inflation cycle, the picture is so far much clearer. The upturn and decline for the category with low import content is at least so far clearly visible. There is also a clear lag and a slower decline so far than for the high import content category.

Figure 6. Figure 6. Average of 2003 and 2008 inflation cycles for CPIF and product categories with high and low import content Annual percentage change

Figure 7. Figure 7. The latest inflation cycle, here from 2021Q1 onwards, for CPIF and product categories with high and low import content Annual percentage change

How much faster does inflation normally fall according to this calculation method for the categories with high import content compared to those with low import content and the CPIF? Figure 8 shows how fast these components in the CPIF decline on average after four quarters from the peaks of the various indices during the 2003 and 2008 cycles. Again, the declines in categories with low import content are followed by increases that are also marginal.

The timing of the various individual peaks has, as mentioned above, been taken into account here. A comparison is made with the latest cycle in Figure 9, where the calculation is for two quarters as data are only available for two quarters after the peak for a number of sub-indices, namely until the first quarter of 2024.

Figure 8. Figure 8. Decline in post-peak inflation for the CPIF and product categories with high (blue bars) and low import content (red bars), average of 2003 and 2008 cycles Number of percentage points

Figure 9. Figure 9. The latest inflation cycle. Decline in post-peak inflation after 2022Q4 for the CPIF and product categories with high (blue bars) and low import content (red bars) Number of percentage points

According to this comparison, the decline in energy-related sub-indices such as “Electricity” and “Fuels” appears to be significantly faster than in other components. The result for "Electricity" is at odds with the result in Figure 5. The “Rents” and “Other services” sub-indices, according to this method, seem to be normally more persistent in terms of the rate of price increase than other components, although for “Other services”, it has fallen faster than normal over the latest cycle.