• No results found

This section investigates the pass-through of changes in costs of fi nal goods for sale (using import and producer (PPI) prices), specifi cally items of HICP non-energy industrial goods.137 The pass-through of the import prices of fi nal goods and of PPI for domestic sales to non-energy industrial goods prices is examined in an autoregressive distributed lag (ARDL) model. As this approach faces some limitations, the pass-through of total import prices is looked at, following Francoise et al. (2007, 2008).

In the following analysis, import prices are measured by unit value indices (UVI) for intra and extra-euro area imports.138 UVI and PPI data have been mapped for 15 product sectors with HICP data, covering about 23% of the HICP and about 80% of non-energy industrial goods for the euro area. The mapping table is shown as Table A15 in the Technical Appendix.139 Pass-through of import and producer prices to consumer prices

The pass-through of domestic and foreign costs is analysed fi rst, using ARDL specifi cations.

This means that domestic consumer prices are

While it would also be interesting to have information on the 137

impact of the structural features on the pass-through of other costs, in particular wages (on a sectoral level), which are of high relevance for the retail sectors, the data needed for this analysis are not available.

Imports are based on the CPA 2002 trade data, while 138

industrial producer prices (PPI) are based on the NACE Rev.

2 classifi cation.

Importantly, this mapping of different classifi cation schemes 139

has its caveats, as, in many cases, the UVI, PPI and HICP cover not identical goods, but just a similar class of goods.

Moreover, while the PPI and HICP are “real” price indices with a well-defi ned basket of goods and high statistical standards, UVI are usually of lower data quality.

explained by their own lagged values and the current and lagged values of all other main explanatory variables:

Variables in lower case letters are in logs, while fi rst differences account for unit roots in the time series.140 The series are quarterly, the longest covering the period from the fi rst quarter of 1990 and the third quarter of 2010, but in many cases the estimation period is shorter, depending on the availability of data.141 Owing to data limitations, only ten euro area countries are covered in this analysis: Belgium, Germany, France, Ireland, Italy, Spain, the Netherlands, Austria, Portugal and Finland. Estimation results are shown in Table A16 in the Technical Appendix, where only positive pass-through coeffi cients signifi cant at the 5% level are reported.

Looking at the PPI pass-through in terms of the median estimate of a product category across countries, it appears that a pass-through of more than 0.5 (in descending order) is found for “Jewellery”, “Personal transport equipment”, “Information processing equipment”, “Equipment for reception (TV/

radio)”, “Cars”, “Textiles” and “Furniture”.

An almost complete or full pass-through was found for “Jewellery” in quite a number of countries. A pass-through for the PPI below 0.5 was found for “Pharmaceutical products”,

“Personal care appliances”, “Newspapers/Books”,

“Non-durable household goods”, “Sports equipment” and “Household appliances”.

Meanwhile, UVI pass-through estimates are considerably lower, with only one sector (photographic equipment) having a pass-through estimate exceeding 0.5.

Overall, price changes for domestic goods (i.e. PPIs) tend to be of higher importance for prices of manufactured consumer goods in the larger euro area countries, refl ecting signifi cant domestic production, while import price changes (i.e. UVI) are more relevant for

consumer prices in smaller, more “open” euro area countries where imports play a greater role.

The latter seem to be linked to an import content that is likely to be higher in retail sales and to very little own production, which is also partly a reason for the reduced availability of PPI data for these countries.

However, no signifi cant association is found between these pass-through estimates across products/countries and structural features of the retail sector – measured by the HHI and the profi t share. This fi nding does not necessarily mean that the pass-through of costs is independent of structural features in the retail sector in the euro area economies, but rather illustrates that the analysis faces many diffi culties, partly related to the availability and quality of price and cost data, as well as that of structural indicators, at a detailed sector level.

Impact of import prices on consumer prices Given that there were some limitations to the previous ARDL approach, a simpler analysis is now applied to the impact of import prices on consumer prices. Mainly following the approach of Francoise et al. (2007, 2008), domestic producer prices are approximated by intra-area import prices.142 More precisely, long-run pass-through elasticities of non-energy industrial goods prices to changes in total import prices are estimated using the following equation:

ln HICPt = α + β ln UVIt (+δT)

Dummy variables for changes in the country’s standard VAT 140

rate are included if signifi cant. The lag structures are determined by reducing a general specifi cation to a parsimonious one by F-tests and t-tests on the signifi cance of sets and single parameter estimates. The long-run elasticity of the pass-through of producer prices to domestic consumer prices is given by

i=Lp

i=Lh

γi / (1– βi).

Consumer prices of clothing and footwear are seasonally adjusted, 141

taking account of a strong and changing seasonal pattern.

The other variables are not adjusted, as the autoregressive part of the equation is able to capture the seasonality adequately.

While this is a rather bold assumption, it allows use of a 142

consistent data classifi cation set to be made, a high level of coverage across euro area countries and sectors, and the extraction of just one pass-through estimator.

STRUCTURAL FEATURES ON PRICE LEVELS, PRICE-SETTING BEHAVIOUR, REGIONAL PRICE DYNAMICS AND PASS-THROUGH where the log of HICP prices is regressed on a

constant and the log of the respective UVI (intra and extra-area import prices). As prices of a number of electronic products have been on a downward trend due to technical progress and related quality adjustment in the HICP, which is not suffi ciently refl ected in UVI, a time trend is added to the equation.143 Only the estimated parameters which are positive and signifi cant at the 5% level are reported. 144

The median elasticity of import price changes to consumer prices is estimated at 0.45 across the 12 euro area countries and 15 sectors considered in this analysis (see Table A17 in the Technical Appendix). However, the pass-through across industries and countries is rather dispersed. For comparison reasons, the estimated elasticities for the euro area as a whole are added, with an estimated median pass-through of 0.54. This is largely comparable with results in the literature, such as those of Hahn (2003), who estimates the pass-through of non-oil import prices to the overall euro area HICP at 0.31 after three years.

It is possible that this latter estimate is kept down by the very low pass-through of import prices to consumer energy prices, which are included in the overall HICP.

In ten sectors, the median long-run import price elasticity of consumer prices is around 0.5, and, in most of these sectors, import price elasticities are signifi cant for almost all countries. This is particularly true of the sectors

“Furniture”, “Appliances for personal care”,

“Jewellery, clocks and watches”, “Newspapers, books and stationery” and “Motor cars”.

Fewer signifi cant results at the country level are found in the clothing and footwear sector, where the median elasticity is 0.4, but with large country dispersion. The same applies to the items “Information processing equipment”

and, albeit to a lesser extent, “Photographic equipment, etc.” and “Household appliances”.

The inclusion of a time trend in the estimates for these two sectors, which attempts to capture the impact of technical progress, does not help in detecting a signifi cant relationship between the HICP and UVI. This may also explain the few

signifi cant and meaningful results at the country level for the “Household appliances” sector, with a median pass-through of 0.1. There is also little discernible pass-through in the “Games, toys and hobbies” and “Equipment for sports, camping and open-air recreation” sectors.

Regarding country divergence, consumer prices seem to have relatively high import price elasticities (i.e. a median higher than 0.5) in Belgium, the Netherlands and Spain, and a relatively low one (median of 0.2) in Ireland.

Overall, the fi ndings are somewhat different to those of Francoise et al. (2008), but these are based on a different sample with respect to the countries, period and goods. Nevertheless, they confi rm that the impact of import prices on consumer prices varies greatly across countries and sectors.

There is some evidence that the magnitude of the estimated pass-though is related to the degree of competition/concentration in the specifi c country and sector. Chart 17 shows a negative, albeit weak, relationship between the estimated import price elasticity and the HHI, suggesting that the stronger the competition (i.e. the lower the HHI), the higher the elasticity of consumer prices seems to be with respect to import price changes. By contrast, no link seems to exist between the estimated coeffi cients and the profi t share (see Chart 18).

To check the robustness of the above relationship, an estimation is made of whether competition signifi cantly impacts on the magnitude of the import price elasticity, once controlling for effects stemming from cross-sector differences. A panel regression

All series have been seasonally adjusted by ARIMA X12. SUR 143

estimation is applied to allow for contemporaneous correlation between the error terms across the country equations for a specifi c sector, as the shocks are expected to be sector rather than country-specifi c.

UVI and HICP series are, in most cases, non-stationary, which 144

implies that conventionally used tests do not have standard asymptotic distributions. However, the HICP and UVI series are not expected to be co-integrated, as an important part in this relationship, namely the costs of domestically produced goods, is missing.

with sector fi xed effects is run, where the import price elasticity is explained by the HHI. As can be seen in Table 22, the measure of competition used here has some explanatory power for the import price elasticity, confi rming that competition leads to a higher transmission of cost changes, although the overall impact is not very strong. This fi nding is in line with theory suggesting that transmission of cost changes is complete in perfectly competitive markets and similar to that obtained in Francoise et al. (2008). However, other structural indicators, such as the profi t share and the OECD product

market indicator, do not help to explain the observed differences in pass-through estimates.

Chart 17 Estimated import price elasticity