Price Setting Behaviour in Denmark

Bo William Hansen and Niels Lynggård Hansen, Economics

INTRODUCTION AND SUMMARY

For many years, prices in Denmark have shown a relatively stable development. Since 1990, average annual inflation has been slightly above 2 per cent. In a well-functioning market economy, average price stability by no means entails that all prices are stable, however. On the contrary, rising or falling prices signal which goods are in short, or ample, supply. The more stable prices are on average, the stronger the signal effect of individual price changes. This is why Danmarks Nationalbank, like other central banks, seeks to keep prices stable on average, but is not in favour of controlling individual prices, since control prevents the price mechanism from signalling a shortage or a surplus. Frequent price changes can thus be perceived as a sign of health.  

On average, 17.3 per cent of all Danish consumer prices are adjusted in a given month. This is shown by a study of the detailed prices collected by Statistics Denmark for calculation of the Danish CPI, cf. Hansen and Hansen (2006). This is around the same adjustment frequency as in the euro area, while prices in the USA are more flexible.  

These results are by no means controversial in relation to modern macroeconomic theory and thinking, where price stickiness is a standard assumption that plays an important role in the inflation process and for how an economy reacts when exposed to a wide range of shocks. For example, monetary shocks such as changes in interest and exchange rates will, all other things being equal, have a greater impact on economic activity when prices are adjusted slowly. Despite recent decades' extensive theoretical literature on price stickiness at micro level, and its significance to e.g. the functioning of monetary policy, there have been surprisingly few empirical studies of micro data for price setting behaviour. Most of these have been based on a limited number of products and price observations. Bils and Klenow (2004) are the first to systematically analyse the detailed price information behind the US CPI, and thereby to cover a substantial proportion of the households' expenditure basket. Equivalent analyses have subsequently been conducted for a number of other countries. This article describes the first analysis of this type of the detailed price data behind the Danish CPI.

The applied data covers the period 1997-2005 and comprises around 2.7 million monthly price observations. The material reveals considerable variations across sectors and products in terms of the frequency and size of the price changes. The prices of unprocessed food and energy are clearly the most flexible, with price changes for, on average, around half of all products in the course of a month. Only around 10 per cent of prices for processed food, industrial goods and services are changed every month. Almost 40 per cent of price changes are a price decrease, so that prices are not in general more rigid downwards than upwards. The prices of a number of products are adjusted at regular intervals or in typical months, e.g. in connection with bargain sales, while prices of other products do not follow any clear timing pattern. Generally, the price changes are high compared with the inflation rates that can be observed from the aggregate price indices, and the price decreases are typically a little higher than the price increases.

The significance of these results to the inflation process is analysed by simple correlation analyses. They show aggregate inflation to be positively (negatively) correlated with the frequency of price increases (price decreases), while inflation does not seem to be correlated with the size of the price changes.

DATA AND METHOD

On a monthly basis, Statistics Denmark collects around 25,000 prices in order to calculate the Danish CPI. Most of the collected prices can be attributed to a specific product in a specific retail outlet at a certain time, but due to the statistical confidentiality restriction neither the names nor locations of the outlets can be identified.

Chart 1 presents examples of individual price series, selected for their typical patterns. The price setting behaviour clearly varies substantially, and the average length of the period between two price changes – also called the duration of a price spell – varies considerably between products. The price of bananas is changed frequently, normally at least once a month. On the other hand, prices of other types of food vary far less frequently, as illustrated by the price of salt. The frequency and size of price changes are to a high degree related to the type of product. Bananas – in contrast to e.g. salt – cannot be stored for a long period, and their prices may be changed frequently in order to sell them before their " sell-by date" is reached. The price history for electric kettles is another example of how certain prices are infrequently changed, but also of significant adjustment of the price level when e.g. a new model appears in the shops, as well as of temporary reductions in a bargain sale. Likewise, the prices of many types of services, in this case a man's haircut, are infrequently changed, and often at certain times of the year, if they are.

The quality of the dataset is generally high and well-suited for our analysis. For various reasons, just over 100,000 price observations were eliminated, the majority being administered prices, i.e. prices that are subject to a high degree of government regulation (e.g. public transport and childcare). This is because we are only interested in analysing the market-determined price setting.

In order to make the study comparable with similar studies in other countries the data is divided into five relatively homogenous product groups, namely 1) unprocessed food, 2) processed food including alcohol and tobacco, 3) energy, 4) non-energy industrial products, and 5) services.

The analysis focuses on the frequency and size of the price changes and their covariation with aggregate inflation. For this purpose, aggregate statistical measures are set up, whereby respectively the frequency and size of the price changes are first compiled as simple averages for each product category; then the aggregates are calculated as weighted averages by applying official HICP weights published by Statistics Denmark. At the most detailed level, around 450 product categories are individually weighted.

The filtered dataset constitutes around 85 per cent of HICP's weight basis, cf. Table 1. Especially the weight of services has been reduced, which reflects that most of the administered prices lie within this group. In terms of observations, unprocessed food is clearly overrepresented, representing 25 per cent of all observations in the dataset, despite having a weight of only 7 per cent (6.0/85.2).

HOW OFTEN ARE PRICES ADJUSTED?

On average, 17.3 per cent of prices are adjusted in a given month – 10.2 per cent of prices are raised, and 7.1 per cent are lowered, cf. Chart 2. This means that around 40 per cent of all price changes are reductions, so that prices are not in general more rigid downwards than upwards.

The frequency of price increases and reductions varies considerably across product categories. The prices of energy and unprocessed food are clearly the most flexible. Overall, more than half of all energy prices are changed monthly. This confirms that these two components are also the most volatile at aggregate level, and they are typically omitted from the calculation of core inflation. Energy is composed of items of which prices are changed several times in the course of a month, e.g. fuel at the petrol station, and also other items of which prices are changed infrequently. One example is district heating, of which the price is typically adjusted once a year. For unprocessed food, especially fruit and vegetable prices are very flexible. For example, 92.7 per cent of the collected banana prices are changed on a monthly basis, cf. above.

The greatest price stickiness is observed for services, closely followed by processed food and industrial goods. Only 2.2 per cent of prices for services are lowered per month on average, equivalent to one fourth of all price changes for services. Considering the underlying service items, it is seen that especially restaurant prices are rarely lowered. Only 0.2 per cent of prices for a cup of coffee are reduced from one month to the next. This pattern presumably reflects that the production of services is very labour-intensive, so that stickiness in wage development is reflected in price setting.

To a degree, the ratio of prices changed in a given month varies over time, and the price movements for some products seem to adhere to a seasonal pattern. This e.g. applies to many unprocessed foods that are sensitive to weather conditions and seasonal variations, just as the prices of a number of industrial goods, e.g. clothing, are influenced by regular bargain sales. For other products such as insurance, district heating or electricity, prices are only changed at particular times. The ratio of prices that are changed in a given month may also be affected by changes in indirect taxes. This was clearly apparent in connection with the cross-frontier-trade package that entered into force in Denmark on 1 October 2003, where reductions of excise duties on alcohol and cigarettes led to strong price decreases for these products in the same month.

THE SIZE OF PRICE CHANGES

Chart 3 shows that, on average, price changes are considerable.^[1] The typical price increase is by 12.3 per cent, and the size of price decreases is even higher, at 15.6 per cent. This clearly exceeds the inflation rates that can be observed from the aggregate price indices.

Price changes for unprocessed food are the highest, which is not surprising since these are perishable goods of which the supply is influenced by the weather and the seasons. Moreover, the price elasticity of these foods is probably small: if a batch of bananas, for example, is to be sold before the end of the week, it can be necessary to reduce the price considerably. The smallest price changes are for energy, which reflects both the frequent, moderate price changes for fuel at petrol stations, and the annual adjustments of prices for e.g. district heating and electricity.

Chart 4 shows the distribution of the price changes by size. The largest share of the distribution is placed to the left of the zero point, since, as stated above, prices are slightly more often raised than lowered. This is especially apparent for services, which account for the smallest share of price decreases.

There are obviously significant variations in the distributions across product groups. For unprocessed food, the distribution is almost symmetrical around the zero point, with many large price changes, and virtually no price changes below 5 per cent. For energy, almost all price changes are below 10 per cent, while price changes for industrial goods show greater dispersion, with a certain proportion of very large price adjustments.

Unlike frequency, the size of price changes does not show a clear seasonal pattern. The clearest sign of seasonal variation is for industrial goods, with summer and winter sales in the retail sector leading to large price decreases.

WHAT DRIVES INFLATION – THE FREQUENCY OR THE SIZE OF PRICE CHANGES?

There is considerable variation in both the frequency and the size of price changes. On this basis, it is interesting to investigate whether there is a connection between these micro-based observations and the development in aggregate inflation.

To gain a first-hand impression of the covariation between aggregate inflation and the frequency and size of price changes, we compare year-on-year inflation rates with 12-month moving averages of the frequency and size of price increases and reductions, cf. Chart 5.

The increase in overall inflation from 1 to 3 per cent in 1999 was clearly related to a higher frequency of price increases and a lower frequency of price decreases. This indicates that the increase in inflation in 1999 was driven by a larger relative price-increase ratio. On the other hand, the development in the size of the price changes does not contribute to explaining the upturn in the inflation rate, since the size of price increases during the same year diminished slightly, while the size of price decreases rose. The same applies to the development in inflation from 2003 to 2005. In this case, the decrease in inflation in 2003 appears to be related to a lower ratio of price increases to price decreases, while the size of price increases and decreases over the same period again took the " wrong" course in terms of explaining the development in inflation. The tendency for inflation to be driven by the frequency, and not the size, of price changes becomes even more apparent when the individual product categories are considered, since the frequencies and sizes calculated for the overall dataset conceal considerable heterogeneity.

To obtain a more accurate picture of this covariation, we consider cross plots of month-on-month aggregate price-increase rates by product categories on the one hand, and respectively the frequency and size of price increases/decreases on the other. In general, this confirms that the frequency of both price increases and decreases is correlated with the month-on-month inflation rates, and with the correct sign. On the other hand, the correlations between size and month-on-month inflation are very small. In Chart 6 this is shown for unprocessed food, where the correlation is most apparent, but it also applies to the other product categories, with non-energy industrial goods as an exception.

PRICE SETTING BEHAVIOUR IN DENMARK COMPARED TO THE EURO AREA AND THE USA

Price setting in the USA and the euro area is analysed by respectively Bils and Klenow (2004) and Dhyne et al. (2005).^[2] The period studied in the two analyses does not completely match ours, but all studies refer to periods of low inflation.

Chart 7 shows the frequency of price changes in Denmark, the euro area and the USA across product categories. In order to make the differences within the categories where prices are changed infrequently more apparent, the volatile groups of unprocessed food and energy are excluded. As stated, these two components are typically eliminated from the compilation of core inflation. In principle, there is no particular reason to believe that the price setting for perishable food products differs significantly from country to country, and nor is this the case. On the other hand, when comparing price setting in the energy sector across countries it should be borne in mind that while the price of fuel varies considerably in most countries, there can be large differences in, for example, the length of the price contracts offered to consumers for deliveries of electricity, gas and district heating. In addition, there can be great variation among countries in the ratio of homes with district heating or oil-burning furnaces.

The frequency of price changes for processed food and services is slightly higher in Denmark than in the euro area, while the frequency of price changes for industrial goods is marginally lower in Denmark. In the USA, the prices in the three categories are adjusted a good deal more frequently; for example, the frequency of monthly price changes for industrial goods and services is more than twice as high as in the other two areas. This is a remarkable difference which cannot be explained solely by the fact that the aggregate monthly inflation rates in the USA were also generally a little higher in the period analysed. The stickier European consumer prices may reflect a higher degree of regulation of labour and product markets than in the USA. If the development in European wages and other costs is stickier, this can contribute to more rigid price setting.

LITERATURE

Bils, M. and P. J. Klenow (2004), Some Evidence on the Importance of Sticky Prices, Journal of Political Economy, vol. 112, no. 5, pp. 947-985.

Dhyne, E., L. J. Alvarez, H. Le Bihan, G. Veronese, D. Dias, J. Hoffmann, N. Jonker, P. Lünnemann, F. Rumler and J. Vilmunen (2005), Price Setting in the Euro Area: Some Stylised Facts from Individual Consumer Price Data, ECB Working Paper no. 524.

Hansen, B. W. and N. L. Hansen (2006), Price Setting Behaviour in Denmark – A Study of CPI Micro Data 1997-2005, Danmarks Nationalbank, Working Paper no. 39.