The Cyclicality of Domestic Prices

Bo William Hansen and Dan Knudsen, Economics

INTRODUCTION

Since 1984 Danmarks Nationalbank has on a regular basis published an index of domestic market-determined inflation (IMI), i.e. the overall consumer-price index stripped of energy, food, administered prices and imports. Compilation of the IMI price index is described by Hansen and Knudsen (2005).

Inflation is the persistent increase in the general level of prices in an economy. The word " persistent" indicates that inflation cannot merely be measured as the year-on-year increase in the consumer price index. Various efforts are therefore made to rinse the consumer price index of random fluctuations so that the underlying trend in price increases becomes more visible. Energy and food are two categories where prices fluctuate substantially, and one approach is to eliminate these two categories from the consumer price index. The increase in this reduced
index is referred to as core inflation. A more general approach is to eliminate or attach a low weight to any good or service for which the price has risen or fallen significantly.

Like the index of core inflation, the IMI index eliminates energy and food, but it is important to stress that the IMI index is not aimed at capturing the underlying inflation trend. The primary purpose is to capture the part of the price development that is determined in a Danish
market^[1]. Decisive to the usefulness of the IMI index is therefore not whether it shows a stable trend, i.e. price increases with low volatility, but that it can be related to economic development in Denmark.

This article views the IMI index in relation to cyclical fluctuations and to other price and wage indices, while also discussing whether IMI can be seen as an indicator of future inflation. In addition, delineation of the IMI index is discussed in the Appendix.

THE PRICE OF VALUE ADDED

The IMI index relates to the price of value added. The method applied when compiling the IMI index is to strip the overall consumer price index of components seen as " exogenous" to price formation in the domestic market. These are energy, food (including beverages and tobacco), administered prices and indirect taxes. In addition, the price impact from the import content in the remaining goods and services is eliminated in the IMI index, as is a small contribution from the indirect energy content, cf. Hansen and Knudsen (2005) on the calculation method. The calculation from consumer price index to IMI index is shown in Table 1.

The aggregate eliminated items have considerable weight in private consumption, so that only just under 35 per cent of total price formation is attributable to the Danish market. It is a point of discussion what to eliminate from the overall price index in order to obtain the best index of the domestic market-determined price. Most goods categories contain both exogenous and domestic market-determined price elements. For example, food is subject to a high degree of domestic processing. The Appendix shows that if processed food and beverages are included, IMI increases to 45 per cent of private consumption, but the movement in the IMI index changes only slightly.

If the impact from fluctuations in import prices is eliminated, the result is the price of value added on supplies for private consumption. The value added is equivalent to production less input of raw materials, and expresses the proceeds available for remuneration of the factors of production. The development in the IMI index thus corresponds to the
development in wages and gross profits on supply of the goods and services included in the IMI index.

The IMI index relates to value added from the manufacture of goods for private consumption, less the value added in energy production, food production and some parts of the public sector. This is a broad definition, which includes large parts of the manufacturing, transport, post, telecommunications and health sectors, as well as hotels and restaurants and a number of enterprises within financing and business service. The building and construction, iron and metal and chemical industries, which primarily supply products for investment and export, are only included to a limited extent. 

We cannot really identify an IMI sector in the national accounts since the value-added deflators of the sectors, e.g. the electronics sector or textile industry, in the national accounts reflect not only supplies for private consumption, but also the price of supplies for investment, export and public consumption. The IMI index approximately resembles the deflator for the private non-agricultural sector^[2], cf. Chart 1. The covariation with an official price index from Statistics Denmark is reassuring since the IMI calculation is based on the monthly consumer price indices and price indices for the domestic supply of goods and is therefore not necessarily close to the prices in the national accounts, which in principle reflect all price statistics.

THE IMI INDEX AND OTHER PRICE AND WAGE INDICES

In many ways the IMI index resembles the overall consumer price index. For example, the average rate of increase in the IMI index has been 4.8 per cent year-on-year since the beginning of 1975, against 4.7 per cent in the consumer price index. In addition, both indices show that inflation has been lower after 1990 than before, cf. Chart 2. However, there are differences; for instance IMI has been fluctuating more than consumer prices in recent years.

In the late 1970s, domestic inflation was substantially below overall inflation, which may have reflected the weak economy with high unemployment. In the mid-1980s the economy was booming, and domestic inflation was relatively high, while wages were accelerating, cf. Chart 3. The most recent gaps between the IMI index and the consumer price index were seen in 2000 and notably in 2005 when the IMI index actually fell year-on-year. This may reflect that the capacity pressure in 2005 was too moderate to allow price rises for imported goods and energy to be passed on to consumers. In 2006, the IMI index has begun to accelerate, cf. Recent Economic and Monetary Trends.

In connection with the above episodes there tends to be a negative correlation between increases in the consumer price index and the IMI index. Particularly import prices separate the two indices. Thus, the development in the consumer price index is closer to the development in the IMI index before import prices are eliminated.

 As described above, the IMI index resembles a value-added deflator, and therefore payroll costs are a significant explanatory factor behind the development in the IMI index. In addition, the sluggishness in price formation means that the development in import and energy prices is reflected in the IMI index. More specifically, the IMI index drops in connection with import price hikes^[3]. The same applies to a value-added deflator if the rise in import prices is not immediately passed on fully to product prices.

COVARIATION BETWEEN IMI AND CYCLICAL FLUCTUATIONS

Since the IMI index comprises goods and services for which Danish market conditions determine the price formation, we would expect the IMI index to be more sensitive to domestic cyclical trends than e.g. the consumer price index. Moreover, we expect the reaction pattern of a price index for value added to differ from that of a price index for consumer goods and services.

In order to investigate this further we will look into the covariation between GDP^[4] at constant prices on the one hand and the level of various price and wage indices on the other hand. The period under review is 1987-2002 since Denmark's economic policy was restructured in the early 1980s. All series are seasonally adjusted quarterly data, i.e. a total of 64 observations.

To start with, each of the quarterly series is broken down into a cyclicalcomponent and a trend^[5]. Both GDP and price and wage indices have a pronouncedincreasing trend. Then the cyclical development is studied by viewing the correlation pattern for the cyclical components. The simple correlations between cyclical components describe the pattern of the covariation between variables for real activity and nominal variables.

Chart 4 shows the correlation between the cyclical GDP component and the various indices with a lead of up to 12 quarters.

The correlation between GDP and the consumer price index is negative from the outset and remains so for some quarters. Six quarters later, for example, the correlation between GDP and the consumer price index is still marginally negative. Only after eight quarters does this correlation increase notably. In other words, it takes some time before an inflationary effect of high GDP filters through to the overall consumer price. Cyclical developments in consumer prices are one of the controversial topics in the study of business cycles, cf. Kydland and Prescott (1990), who believed that the negative correlation showed that the business cycles were mainly created by supply shocks.

In contrast, a positive correlation between GDP and the IMI index is seen at an earlier stage. This could indicate that the IMI index reflects activity pressure more rapidly. In the first quarters, however, the correlation is negative, and sudden hikes or dives in GDP cannot be expected to be reflected in the IMI index immediately.

In the calculation of the transition from the consumer price index to the IMI index, particularly the subtraction of the import content sustains the correlation between GDP and the IMI index. This is reflected in the fact that an interim result where the import content has not yet been subtracted correlates negatively with GDP in a way that is closer to the GDP correlation with the consumer price index. The clear difference in the GDP correlations for, respectively, the IMI index and the IMI index including imports reflects how high import prices, i.e. poor terms of trade, have typically been linked to both low GDP and a low IMI index.

Hourly wages can be seen as a domestic market-determined price formed in the Danish labour market. High cyclical GDP means that activity is high. This increases pressure on the labour market, so that a lagged positive correlation between GDP and wages should be seen. In the period reviewed, the correlation between GDP and hourly wages in manufacturing is indeed positive with a delay of four quarters. The effect from GDP on wages greatly resembles the effect from GDP on the IMI index.

Instead of GDP, unemployment can be used to describe the state of the economy. Chart 5 shows the correlation between the cyclical component of unemployment and the nominal variables.

If the impact on wages goes via the labour market, there must be a clearer and more rapid cyclical effect in the correlation pattern between the cyclical components of unemployment and wages than between GDP and wages. It is in fact seen that wages immediately have a negative correlation with unemployment. The IMI index's correlation with unemployment is weaker than the wage correlation, but the development over time resembles the wage correlation with unemployment.

On the other hand, the correlation of the consumer price index with unemployment is visibly slower and weaker, and again we see that the pattern of the IMI index including imports resembles that of the consumer price index. The key information element in the compilation of the IMI index therefore seems to be elimination of import prices.

THE IMI INDEX AS AN INDICATOR OF FUTURE INFLATION?

If the IMI index reacts faster to cyclical developments than the consumer price index, this indicates that the IMI index, like wages, can predict the direction of overall consumer prices.

A simple way to examine the predictive properties of IMI is to relate the current year-on-year increase in the consumer price index and the IMI index to the realised future consumer price inflation:

Applying monthly data over a horizon of e.g. 12 months (h=12), the left side of the equation shows the year-on-year rate of increase in the consumer price index one year ahead. On the right side of the equation, many other explanatory variables can potentially be useful for predicting inflation.

As a benchmark for our results, IMI is replaced in the equation by core inflation defined as the consumer price index excluding energy and food. Core inflation is typically less volatile than both the overall increase in consumer prices and IMI, cf. Chart 6, and internationally it is often used to predict fluctuations in the consumer price index.

Table 2 presents a number of regression results for h=12 and h=24. The data period is January 1985 to September 2006.

With a horizon of 12 months, both core inflation and IMI contribute to predicting consumer price inflation. In both cases, the estimated coefficient is significant, and the degree of explanation is slightly higher when IMI is included in the equation. At h=24, the degree of explanation declines, particularly for core inflation. At the same time we see that only the coefficient of IMI is significant, and thus the regression analysis indicates that a rise in IMI is often followed by a higher overall rate of increase in consumer prices. In other words, the development in IMI may contribute to predicting the development in consumer price inflation two years ahead, while core inflation does not have a similar explanatory power.

CONCLUSION

Our analysis shows that the IMI index is more exposed to cyclical fluctuations in Denmark than the overall consumer price index. The IMI index responds faster than the consumer price index to GDP and unemployment and in this respect resembles the index of hourly wages. In spite of its higher volatility, the year-on-year increase in the IMI index is a better indicator of overall consumer price inflation than core inflation is, particularly with a horizon of more than one year.

The acceleration in the IMI index in 2006 reflects, inter alia, the tight labour market, and the rising domestic inflation pressure indicated by the acceleration in the IMI index will have an impact on overall inflation in 2007.

LITERATURE

Bank of England (1998), Inflation Report, August.

Bank of England (2006), Minutes of The Monetary Policy Committee Meeting 5 & 6 July 2006.

Danmarks Nationalbank (2003), MONA – a quarterly model of the
Danish economy.

Hansen, B. W. and D. Knudsen (2005), Domestic Market-Determined Inflation, Danmarks Nationalbank, Monetary Review, 4th Quarter.

Hodrick, R. J. and E. C. Prescott (1997), Postwar U.S. Business Cycles: An Empirical Investigation, Journal of Money, Credit and Banking, Vol. 29, No. 1.

Kydland, F. E. and E. C. Prescott (1990), Business Cycles, Real Facts and a Monetary Myth, Federal Reserve Bank of Minneapolis Quarterly Review, Vol. 14, No. 2.

APPENDIX: A BROADER DELINEATION OF IMI

All food and beverages are subtracted in the calculation of IMI, cf. Hansen and Knudsen (2005). IMI is thus in line with the Eurostat expression of core inflation in which energy, food, and beverages and tobacco are eliminated from the HICP index.

However, Eurostat also publishes another expression of core inflation, defined as HICP excluding energy and unprocessed food. The European Central Bank, ECB, often uses this expression in its analyses. If the calculation of IMI is linked to this expression of core inflation, the IMI index must be extended to include beverages and tobacco, and processed food.

In the most recent weight basis for the index of net retail prices^[6], food and beverages, including alcohol and tobacco, constitute 14.4 per cent of consumer spending, while unprocessed food (meat, fish, fruit and vegetables) only constitutes 5.1 per cent. The weight basis for IMI is thus increased by 9.3 (14.4-5.1) percentage points to 60 per cent if only unprocessed food is eliminated. Chart 7 shows the increase in two stripped indices of net retail prices in the period January 2001 to July 2006. In spite of their different content, the two indices show a considerable degree of covariation.

The consumption components (IMI products) represented in the stripped index of net retail prices are identified in Statistics Denmark's input-output table for 2000. At the most detailed level, private consumption is broken down into 73 components. If processed food and beverages and tobacco are included, the IMI sector increases from 41 to 55 of these consumption components, but in spite of the different composition, the input-output weights do not change materially.

Import prices are measured using price indices from the statistics for the domestic supply of goods. " Processed food" and " beverages and tobacco" are represented by the " raw materials for agriculture" and " beverages and tobacco" import price indices. Processed consumer goods (e.g. dairy products) are also included in the " foods of animal origin, excl. fish" import price index, but since this index is dominated by a number of unprocessed meat products, it is of no avail. The " new" import price indices are co-weighed with the import price indices used so far, " raw materials for other industries" and " other consumer goods" . The indices are co-weighed in accordance with the weight distribution (from 2000) in the price index for the domestic supply of goods:

Import price index capital goods = (26.1*" raw materials for other industries" + 1.6*" raw materials for agriculture" )/(26.1+1.6)

Import price index consumer goods = (12.3*" other consumer goods" + 0.7*" beverages and tobacco" )/(12.3+0.7)

Since the weights of the two new indices are relatively modest (1.6 and 0.7 per cent, respectively), the import price indices do not change significantly, cf. Chart 8.

Neither the stripped index of net retail prices, the input-output weights nor the import price indices change materially if processed food and beverages and tobacco are included in the calculation of IMI. Even though the IMI products' total share of consumer spending increases from just under 35 per cent to around 45 per cent, the impact from the exercise on the IMI index is limited, cf. Chart 7.