Evaluating Transmission Prices Between Global Agricultural Markets and Consumers' Food Price Indices in the EU. Technical University of Madrid, Spain

Transcription

1 Scientific paper No. 7A July 2014 Evaluating Transmission Prices Between Global Agricultural Markets and Consumers' Food Price Indices in the EU Sol García-Germán 1, Alberto Garrido 1, Isabel Bardají 1 1 Research Centre for the Management of Agricultural and Environmental Risks (CEIGRAM), Technical University of Madrid, Spain Scientific paper, 7A ULYSSES Understanding and coping with food markets volatility towards more Stable World and EU food SystEmS July, 2014 Seventh Framework Program Project KBBE ULYSSES project has received research funding from the European Commission Project KBBE Any information reflects only the author(s) view and not that from the European Union

2 ULYSSES project assess the literature on prices volatility of food, feed and non-food commodities. It attempt to determine the causes of markets' volatility, identifying the drivers and factors causing markets volatility. Projections for supply shocks, demand changes and climate change impacts on agricultural production are performed to assess the likelihood of more volatile markets. ULYSSES is concerned also about the impact of markets' volatility in the food supply chain in the EU and in developing countries, analysing traditional and new instruments to manage price risks. It also evaluates impacts on households in the EU and developing countries. Results will help the consortium draw policy-relevant conclusions that help the EU define market management strategies within the CAP after 2013 and inform EU s standing in the international context. The project is coordinated by Universidad Politécnica de Madrid. Internet: Authors of this report and contact details Name: Sol García-Germán Partners acronym: UPM (Partner 1) sol.garciagerman@upm.es Name: Alberto Garrido alberto.garrido@upm.es Name: Isabel Bardají isabel.bardaji@upm.es When citing this ULYSSES report, please do so as: García-Germán, S. et al., Evaluating Transmission Prices Between Global Agricultural Markets and Consumers' Food Price Indices in the EU, Scientific paper 7A, ULYSSES project, EU 7th Framework Programme, Project KBBE , 29pp. Disclaimer: This publication has been funded under the ULYSSES project, EU 7th Framework Programme, Project KBBE Any information reflects only the author(s) view and not that from the European Union. Any information reflects only the author(s) view and not that from the Food and Agriculture Organization of United Nations. "The information in this document is provided as is and no guarantee or warranty is given that the information is fit for any particular purpose. The user thereof uses the information at its sole risk and liability." ii

3 Table of contents ABSTRACT INTRODUCTION LITERATURE REVIEW PRICE TRANSMISSION IN THE EUROPEAN UNION MODELLING FRAMEWORK DATA RESULTS CONCLUSIONS REFERENCES iii

4 Abstract The rise of price levels and volatility of world agricultural commodities since was followed by increased and more volatile food price inflation around the world. This has raised concerns about the potential effect of higher prices on consumers, not only in developing countries but also amongst the poorest and most vulnerable households in developed countries. Using error correction models, this paper evaluates the velocity and extent to which world agricultural commodity price movements affect consumer food prices in the 28 Member States in the EU. We consider three types of world food commodity price indices. Results show a long run relationship between world agricultural commodity prices and food consumer prices in over half of the Member States. They present varying long run price transmission elasticities and a slow adjustment of prices. There are differences among the Member States in the responsiveness of their consumer prices to different world price indices, suggesting that there are differences in the structure and efficiency of markets among them. In general countries of the Eurozone have lower transmission elasticities. These should be taken into account to predict impacts of extreme world price volatility and rise in consumers' prices. Keywords: Commodity prices, Consumer food prices, Error Correction Models, Price transmission, European Union 1

5 1 INTRODUCTION The price surges occurred during 2006-mid 2008, 2011 and 2012 and the rise in the volatility of agricultural commodity prices have resulted in the increase of the variability of consumers' food price indices both in developing and developed economies. This increase in consumer food prices has raised concerns about their potential effect on the most vulnerable consumers and households around the world (McCorriston 2012). To evaluate how movements in global agricultural commodity markets affect movements in consumer prices in the developed world, it is fundamental to assess the degree and speed of price transmission between world agricultural commodity and consumers' food prices. The literature suggests that the relationship between world and domestic prices may not be too strong (McCorriston 2012). According to Lloyd et al (2012), the prices of raw agricultural commodities tend to be more volatile than food retail prices. Apergis and Rezitis (2011) state that higher food prices translate into higher inflation. So the link among the two is obvious but will depend on the process of price transmission in which many factors are involved. The impact on consumers in developed economies depends on the extent to which consumer food prices respond to agricultural commodity prices. This transmission is usually incomplete due to several factors, softening consumers' food price instability (Gilbert and Morgan 2010). The structure and efficiency of the food sector of an economy affects also the level and velocity of transmission. Higher or more volatile prices may cause greater welfare losses to those consumers who devote a larger proportion of their income to food. To evaluate how movements in global agricultural commodity markets affect consumers in the developed world, we first have to assess the degree and speed of price transmission between world commodity and consumers' food prices. Price shocks pass through to consumer prices to a given extent (Ferrucci et al 2010). However, the impact of global food price volatility on consumers in developing countries takes place through the direct consumption of staples, whereas richer consumers have a more indirect dependency on agricultural commodities through feed grain to produce meat (Gilbert and Morgan, 2010). This paper evaluates price transmission in 28 EU Member States - the extent and speed to which agricultural commodity price movements affect consumer food prices in the MS of the EU - using Engle and Granger error correction models for each of the 28 MS. This paper 2

6 contributes to the literature in several ways. Firstly, all MS of the EU are being analyzed taking into account not only agricultural commodity prices but also supply and demand shifters. Secondly, three different world commodity price indices are used to capture different impacts on the transmission of prices due to the coverage of commodities and the weighting structure used to compile each of them. Since the main research question is to evaluate how movements in global agricultural commodity markets affect overall EU households' costs of food, we do not focus on the transmission of a particular agricultural commodity to a single food category along a specific supply chain but instead on an aggregate index of consumer food prices. Finally, we test whether national currencies could have a role in explaining the differences both in the varying evolution of consumer food prices and the price transmission among EU's MS. The remainder of the paper is organized as follows. Next section summarises the relevant literature. Section 3 provides some basic data and findings about food price movements in the EU. Section 4 and 5 describe the methods and the data used to evaluate the transmission of prices between changes in world markets' prices and food consumers' prices in the different MS of the EU, respectively. Section 6 presents the results and we discuss the results. Finally, in the last section, we synthesize the conclusions. 2 LITERATURE REVIEW In the face of the global rise of food commodity prices, Lloyd et al (2012) review the developments of food price inflation across the EU. These authors note that in OECD countries general inflation volatility begins to rise from the mid-2000 s onwards. When looking at food price inflation - a part of general inflation - in the EU they see that it is higher and more volatile than non-food price inflation. Figure 1 reports overall and food inflation developments in the EU for the period , showing that food inflation volatility is considerably higher than overall inflation. 3

7 Figure 1. Overall and food inflation developments in the EU HICP Annual rate of change (%) Overall Food Source: EUROSTAT The relationship between agricultural commodity prices and consumer food prices depends on horizontal and vertical price transmission (Ferrucci et al 2012; Lloyd et al 2012). The impact on consumers seems to be limited due to the small percentage of the raw commodities' expenditure in the final retail product, which has undergone a certain level of processing. Despite this, commodity prices represent a large fraction of consumer price of fresh products, meat and dairy (Richards and Pofahl 2009). Horizontal price transmission refers to the co-movement of prices between spatially differentiated markets at the same stage of the supply chain (spatial price transmission) or across different agricultural or non agricultural commodities markets (Esposti and Listorti, 2012). According to these authors, non agricultural markets may concern as well other commodities or financial markets. The transmission of prices across borders does not require physical movement of goods and services, the flow of price information is sufficient (von Braun and Tadesse, 2012). This relates to the degree to which markets are integrated, and in the case of price transmission from world agricultural commodity prices to food consumer prices to the degree to which world and domestic markets are integrated (Lloyd et al 2011). Second, vertical price transmission refers to the price linkages along the supply chain. These are characterized by the magnitude, speed and nature of the adjustments which take place along the supply chain to respond to market shocks produced at different levels of it (Vavra et al 2005). 4

8 Commodity raw material is processed to different degrees until it reaches consumers, adding different services (Lloyd et al 2011). Raw materials typically comprise only a small share of processed food in developed countries, so higher agricultural prices are expected to be reflected in consumer food prices to a much lower extent, given the low and declining share of agricultural raw materials in food production costs (European Commission (EC), 2008). Although the literature on price transmission has focused mostly on producer prices within the value chain instead of consumer prices (Baquedano and Liefert 2014), and in the developing world rather than the developed world, several studies have evaluated the passthrough from agricultural commodity to consumer food prices in advanced economies, including the EU's Member States (MS). Some studies of price transmission between world agricultural commodity prices and consumer food prices have been undertaken in the US. Leibtag (2009) uses autoregressive models to evaluate price transmission from corn, soybean and crude oil price changes to farm and wholesale beef, pork, poultry, egg and milk products' prices and the latter to the corresponding retail prices and price transmission from wheat prices to flour prices and the latter to white bread prices. The author shows that the pass-through from food and energy commodity price variations to farm and wholesale prices ranges from 2 to 41% taking from two to nine months, depending on the food category considered. The pass through from farm and wholesale prices to retail prices ranges from 2 to 18% and takes from one to six months. From these results the author derives that the pass-through from commodity to consumer prices ranges from less than 0.5% to nearly 7% and takes from four to twenty seven months. Richards and Pofahl (2009) examine the pass-through from commodity prices to retail prices of apples and cereals by means of a structural demand model. Authors suggest that when commodity prices rise, wholesale and retail margins decline in the apple sector while only retail margins decline in the cereal sector. Whereas when commodity prices go down, the opposite occurs. This suggests the existence of competitive behavior and the performance of pricing power by retailers. Lastly, Berck et al. (2009) estimate the transmission of corn, wheat and gasoline prices to retail cereals and chicken prices using the Generalized Method of Moments (GMM) Arellano and Bond estimator. They find a relatively high elasticity of cereal to flour of over one and to chicken of 0.3. They attribute the rather high elasticities either to imperfect competition or to data length inadequacy. According to Lloyd et al (2012), food price inflation has been higher and more variable in the EU than in Japan and tends to be more similar to that of the US. These authors note that changes of food price inflation differ significantly across MS in the EU, both in level and 5

9 variability. Most MS in the EU experienced high levels and variability of food price inflation from mid-2007 onwards. Many of the recently acceded MS presented also high and volatile food price inflation over the 1990s and 2000s during the transition period, which distinguishes them from the EU-15 MS. 3 PRICE TRANSMISSION IN THE EUROPEAN UNION Traditionally, prices in international markets of certain commodities largely produced in the EU have been lower and more volatile than those prevailing in the EU (Ferrucci et al 2010). According to these authors this can be largely attributed to the Common Agricultural Policy (CAP), which has traditionally mitigated price transmission of world agricultural commodity shocks to EU domestic prices. But no longer does it, due to the severe rise in world agricultural prices and sometimes also to the decrease in guaranteed prices in the EU (National Bank of Belgium, 2008). Different studies have evaluated the pass-through from agricultural food commodity prices to food consumer prices in advanced economies, showing a rather weak relationship between both series of prices (see for example IMF, 2008a 1 and IMF, 2008b 2 ). On the one hand, some papers review the recent evolution of food price inflation in the EU's MS. These point out the different evolution of food price inflation which has taken place recently across MS in the EU (Bukeviciute et al 2009; Lloyd et al 2013). Bukeviciute et al. (2009) suggest that the higher food price increases in the new MS may be due to macroeconomic factors or differences in the food chain structure among MS. To give an example, they consider that the appreciation of their national currencies could have weakened the increase of consumer food prices in some new MS - Czech Republic, Poland, Romania and Slovakia. Lloyd et al. (2013) explain that these differences take place not only between MS of the EU-15 and the EU-12 but also within MS belonging to the EU-15. They conclude that though price inflation - both in levels and volatility - has increased since IMF (2008a) notes that the transmission from world prices to domestic food inflation has been stronger in Europe s advanced economies since the mid-1990s, nevertheless the transmission to core inflation continues to be reduced for most countries. According to the author, the larger pass-through could be due to the reforms that the CAP has undergone. The impact on food price inflation in some of Europe s emerging countries is larger, estimating that around 10% of the variability in domestic food prices can be due to fluctuations in world food prices. 2 IMF (2008b) shows that the transmission from international to domestic food prices and from domestic food prices into core inflation was higher in emerging economies than in advanced economies and that approximately one half of the shocks to domestic food prices reach core inflation in emerging economies, while less than one quarter passes through in developed countries. This is consistent with the differences in the share of food in consumption baskets and in production costs between developed and developing countries. 6

10 in the EU-15, it has decreased in the EU-12. The new MS transition process, the composition of consumers' food expenditure, the extent of import exposure, the degree of market integration, globalization, exchange rates, agricultural policy and market structure and market power are identified as possible factors explaining these differences. Busicchia (2013) notes that in the EU the surge of international agricultural commodities prices was rapidly followed by rises in food producer and consumer prices but the easing of international commodity prices did not spill over domestic food producer and consumer prices. Both producer and consumer prices took more time to stabilize (around 6 months for producer prices and one year for consumer prices) and at a higher level (10% higher for consumer prices) (Busicchia, 2013). On the other hand, some papers evaluate price transmission from agricultural commodity prices - international or domestic - to consumer food prices in the EU. Bukeviciute et al. (2009) emphasize the differences in price transmission among EU MS between domestic agricultural commodity prices and producer food prices in a first step and producer and consumer food prices in a second step. To do so they use error correction models and simple OLS regression. According to these authors, the pass-through elasticities from domestic agricultural commodity prices to producer food prices range from 0.01 (Portugal) to 0.22 (Poland), while the long term elasticities from producer to consumer food prices lie between 0.1 and They suggest that the higher pass-through between producer and consumer food prices in the new MS might be affected by several factors such as the potential contribution of the rise in indirect taxes and larger energy prices to food prices, a larger share of agricultural input prices to final retail prices or price arbitrage, which might have pushed food prices upward in MS with the lowest price levels. In addition, some evidence of price asymmetries is found in the new MS whereas not in the euro area. Bukeviciute et al. suggested that the euro area's retail sector might be more competitive than that of the new MS and that the differences found among MS suggests market fragmentation in the EU. The National Bank of Belgium (2008) studied inflation and price levels developments in processed food in Belgium and in the Euro Area, indicating that the food price rises in the second half of 2007 were connected to the 2006/2008 increase in food commodity prices but consumer prices displayed a less marked increase than commodities due to the fact that they represent a small part of consumer prices. Bank s researchers used vector autoregressive models (VAR) (see Sims (1980) for origins of this) to describe the dynamic relationship between firstly, international food commodity prices and EU producer and consumer prices for the product categories which increased their prices at the end of

11 milk, cheese and eggs, oils and fats and bread and cereals. And secondly, EU internal market prices and EU producer and consumer prices for the same product categories. They find that the internal market price is the appropriate variable for evaluating the movement in processed food prices in Europe instead of the world market price. Besides, Ferrucci et al. (2012) extend the National Bank of Belgium's analysis and find that there is a more significant and long lasting transmission of agricultural commodity prices to consumer prices in the euro area when using EU's agricultural commodity prices instead of international prices. They argue that the study of the price pass-through from international to domestic prices in the Euro Area has to take into account the distortions introduced by the CAP which are accounted for in EU's agricultural commodity prices. Using VAR models, these authors find that the cumulative elasticity of consumer food prices to domestic agricultural commodity prices is 0.33 after a year. They find that a disaggregate approach shows important differences in the structure of pass-through for the various food categories. Lastly, when testing the extent to which the sharp increases in retail prices were due to the 2006/2008 rise in agricultural commodity prices, they point out the strong influence of EU agricultural commodity prices on consumer food prices. Porqueddu and Venditti (2012) analyze the relationship between agricultural commodity prices and consumer food prices in the Euro Area and test whether consumer food prices respond asymmetrically to agricultural commodity prices. They extend the study of Ferrucci et al. by including the analysis of the price pass-through in Germany, Italy and France to test whether there are significant differences in the price pass-through between MS. It differs from the approaches of the studies by National Bank of Belgium and by Ferrucci et al. in that it does not include producer prices. They find that food prices in the Euro Area are affected by commodity prices and highlight considerable heterogeneity across countries and products. At the country level, consumer food prices are generally more responsive in Germany than in Italy and France. These authors note that a recent study suggests that different features of the distribution sector might play a role in these different patterns between countries (see ECB, 2011). Gabrijelcic et al (2012) assess the extent of pass-through of food commodity prices to consumer food prices in Slovenia using world, EU and Slovenian commodity prices. They show that a larger share of the annual growth of Slovenian food prices in 2010/2011 can be explained by the shock in food commodity prices than in 2007/2008. Davidson et al. (2011) model other factors that may drive consumer food prices in the UK, apart from world agricultural commodity prices, using a cointegrated VAR model. They find 8

12 that the major drivers of UK food inflation are world raw food prices and the dollar/pound exchange rate, while manufacturing costs, unemployment and earnings are less important. Oil prices also affect retail food inflation, but indirectly due to its impact in world agricultural commodity prices. In this case, authors find a link between world commodity prices and retail prices and obtain a long run transmission elasticity of between them. According to theses authors, retail food prices increase by 0.28% following a 10% shock in world agricultural prices lasting for one month. If the shock persists 18 months, retail food prices increase by 2.42%, showing that the effect on domestic retail food inflation depends on the duration of the shocks on world commodity prices. Finally, Apergis and Rezitis (2011) assess the behavior of food price volatility and whether the short-run deviations between relative food prices and specific macroeconomic factors have effects on food price volatility in Greece using GARCH and GARCH-X models. The results show that there is a significant effect of the deviations on the volatility of relative food prices, implying greater uncertainty about future prices and market risks. According to these authors, increased food price volatility can reduce the precision of producers and consumers' forecasts of prices and reduce their welfare. 4 MODELLING FRAMEWORK To evaluate the level of integration and the price transmission between world agricultural commodity price indices and consumer unprocessed food prices in the MS of the EU, we formulate error correction models between both price indices for each MS in the EU. These models are augmented with several exogenous variables - the unemployment rate, the exchange rate and a world crude oil price index. The unemployment rate is introduced in the model as a demand shifter, whereas the exchange rate and the world oil price index are introduced as supply shifters. Starting from an autoregressive distributed lag (ADL) model, a parsimonious ECM is derived. These price transmission models permit us to describe both the long run equilibrium relationships and the short run dynamics between world and consumer prices. They permit estimating both the extent to which movements in agricultural commodity prices affect consumer prices and the time it takes for a shock in commodity prices to be passed through to consumer prices (Baquedano et al 2011). 9

13 It is hypothesized that causality runs from world agricultural commodity prices to consumer unprocessed food prices. Under the assumptions of weak exogeneity between the cointegrating variables, the error correction model seems an appropriate approach to undertake the analysis. It consists of a two-step estimation approach. An initial regression specifies the error correction term (ECM), which forms the vector of the stationary residuals of a linear combination of two or more integrated variables. The fact that the residuals of the first regression are stationary shows that cointegration between both price series exists. According to Engle and Granger (1987), if there is evidence of two or more series being cointegrated, then a valid error correction specification should also exist between the variables. The ECM is added to the second regression as a lagged regressor, defining thus a restricted form of the general autoregressive distributed model (Baquedano and Liefert 2014). Therefore the approach is restricted to non-stationary, integrated of the same order, and cointegrated price series. To test for non-stationary and the order of integration, Augmented Dickey-Fuller (ADF) and Phillips-Perron (PP) stationarity tests are performed in all considered series. Once the non stationarity of the data has been shown, in order to test for cointegration using the Engle-Granger method, we check whether the disequilibrium errors or residuals of the first regression are stationary. According to De Boef and Keele (2008), in time series analysis an exogenous variable may have only short term causal effects on the endogenous variable or may have both short and long run causal effects. When there are only short term effects, the effect does not persist into the future. However, long-run effects are distributed across future time periods (De Boef and Keele, 2008). These authors claim that the behavior of the endogenous variable depends on the exogenous variable in the long run, and short run movements in the outcome variable respond to deviations from the long run equilibrium. The parameter of the lagged regressor of the ECM measures the speed with which the model returns to the long run equilibrium following an exogenous shock. The first stage regression, which defines the ECM or the long run equilibrium relationship between both price series, is specified by the following equation, which in simple notation is specific for one country: cp = α + ϕ + ε t wp t t [1] 10

14 where cp t and wp t are the natural logs of the consumer unprocessed food index and the world agricultural commodity price index respectively, ϕ is the long run price transmission elasticity of a change of consumer unprocessed food prices with respect to world agricultural commodity prices and ε t are the residuals or the disequilibrium errors of the equation. To derive the error correction model or second step regression, we start from the ADL model, which is defined by the following equation: cpt = α 0 + α cp t i + α wp t i + α e t i + α u t i + α o t i + εt i = 1 1 i i= 0 2 i i= 0 3 i i= 0 4 i i= 0 5 i [2] where e t, u t and o t are the natural log of exchange rate, the unemployment rate and the oil price index respectively and ε t is the error term. We can derive an unrestricted error correction model, by adding and subtracting lags of the variables in Equation (2). The unrestricted model is defined by Equation (3). Insignificant lags are eliminated to specify a more parsimonious model. Δcpt = β 0 + β Δcp t i + β Δwp t i + β Δe t i + β Δu t i + β Δo t i + εt i = 1 1 i i= 0 2 i i= 0 3 i i= 0 4 i i= 0 5 i [3] where Δ cpt, Δ wpt, Δ et, Δ ut and Δ ot are the first differences of the natural logs of the consumer unprocessed food index, the world agricultural commodity price index, the exchange rate, the unemployment rate and the world crude oil price index, respectively. Equation (3) can be rewritten to incorporate the long-run relationship in Equation (1). Thus the error correction model is specified by the following equation: Δcpt = β 0 + γ ( cp ϕwp ) + β Δcp t i β wp t i β e t i β u t i β Δo t t i i + Δ + Δ + Δ + i i i i i i i i t i + ε 1 1 = 1 1 = 0 2 = 0 3 = 0 4 = 0 5 t [4] where the expression cp ϕ wp ) is the lagged ECM or the stationary residuals of ( t 1 t 1 Equation (1) and γ is the speed of adjustment to equilibrium of the consumer unprocessed food index after a world commodity price shock. The speed of adjustment to equilibrium is the rate at which consumer unprocessed food prices return to equilibrium after a shock in 11

15 agricultural commodity prices. According to Baquedano and Liefert (2014), cointegration is based not only on the residuals of Equation (1) being stationary, but also on the adjustment parameter being different from zero. 5 DATA Data for the unprocessed food harmonized index of consumer prices (HICP) were obtained from EUROSTAT. The unprocessed food price index includes certain sub indices of the more aggregate food index, namely, those corresponding to meat, fish, fruit and vegetables sub indices. The unprocessed food HICP index is used instead of a more aggregate food HICP index in order to minimize the underestimate of price transmission due to potential processing and retail costs which are not taken into account. Three indices of world agricultural commodity prices are used - one from the International Monetary Fund (IMF) and two compiled by the European Central Bank (ECB). The two ECB's indices differ in that one of them is weighted according to the Euro Area import values and the other one is weighted according to the domestic demand or use in the Euro Area. The world price index compiled by IMF 3 is a weighted average of individual commodity price indices and weights depend on their relative trade volumes of each commodity 4 compared to total world trade. Both ECB's price indices 5 cover the same range of food commodities 6 and their prices are world market prices. Both ECB's indices, which are euro denominated, have been converted to US dollars. Figures 2 to 10 show the evolution of the unprocessed food HICP index for selected groups of MS compared to the evolution of the IMF's world food price index. The different MS have been classified into various groups. Due to the differences both in the evolution of food prices and the transmission of prices among MS which was stated in the previous section, MS were grouped geographically in order to reveal similarities or differences between them. MS have been grouped too according to whether they are traditional MS or new MS. Following this, Figure 2 shows the evolution of the HICP index in Greece, Italy, Portugal and Spain, Figure 3 shows the evolution of the HICP index in Austria, France and Germany, 3 IMF Primary Commodity Prices ( 4 The world price index compiled by IMF includes cereals, vegetable oils, meat, seafood, sugar, bananas and orange price indices 5 ECB Statistical Data Warehouse ( 6 The world price indices compiled by ECB includes barley, maize, rice, wheat, soya beans, sunflower seeds, coconut oil, palm oil, sunflower seeds oil, beef, swine, meat, cocoa, coffee, sugar, tea, tobacco, bananas and oranges. 12

16 Figure 4 shows the evolution of the HICP index in Belgium, Luxembourg and Netherlands, Figure 5 shows the evolution of the HICP index in Ireland and United Kingdom, Figure 6 shows the evolution of the HICP index in Denmark, Finland and Sweden, Figure 7 shows the evolution of the HICP index in Czech Republic, Poland and Slovakia, Figure 8 shows the evolution of the HICP index in Bulgaria, Hungary and Romania, Figure 9 shows the evolution of the HICP index in Estonia, Latvia and Lithuania and Figure 10 shows the evolution of the HICP index in Croatia, Cyprus, Malta and Slovenia. As shown in the figures, in some of the considered groups, the consumer indices follow different trends - for example in the case of the Southern MS or Ireland and UK, the index follow quite different trends, while in the case of other MS - such as Austria, France and Germany -, they seem to follow more similar trends. Figure 2: Evolution of the unprocessed food HICP index for groups of selected MS - Greece, Italy, Portugal and Spain - and of IMF's world agricultural commodity price index Unprocessed food HICP index World price index 2000m1 2002m1 2004m1 2006m1 2008m1 2010m1 2012m1 Greece Italy Portugal Spain World 13

17 Figure 3: Evolution of the unprocessed food HICP index for groups of selected MS - Austria, France and Germany - and of IMF's world agricultural commodity price index Unprocessed food HICP index World price index 2000m1 2002m1 2004m1 2006m1 2008m1 2010m1 2012m1 Austria France Germany World Figure 4: Evolution of the unprocessed food HICP index for groups of selected MS - Belgium, Luxembourg and Netherlands - and of IMF's world agricultural commodity price index Unprocessed food HICP index World price index 2000m1 2002m1 2004m1 2006m1 2008m1 2010m1 2012m1 Belgium Luxembourg Netherlands World 14

18 Figure 5: Evolution of the unprocessed food HICP index for groups of selected MS - Ireland and United Kingdom - and of IMF's world agricultural commodity price index Unprocessed food HICP index World price index 2000m1 2002m1 2004m1 2006m1 2008m1 2010m1 2012m1 Ireland United Kingdom World Figure 6: Evolution of the unprocessed food HICP index for groups of selected MS - Denmark, Finland and Sweden - and of IMF's world agricultural commodity price index Unprocessed food HICP index World price index 2000m1 2002m1 2004m1 2006m1 2008m1 2010m1 2012m1 Denmark Finland Sweden World 15

19 Figure 7: Evolution of the unprocessed food HICP index for groups of selected MS - Czech Republic, Poland and Slovakia - and of IMF's world agricultural commodity price index Unprocessed food HICP index World price index 2000m1 2002m1 2004m1 2006m1 2008m1 2010m1 2012m1 Czech R. Poland Slovakia World Figure 8: Evolution of the unprocessed food HICP index for groups of selected MS - Bulgaria, Hungary and Romania - and of IMF's world agricultural commodity price index Unprocessed food HICP index World price index 2000m1 2002m1 2004m1 2006m1 2008m1 2010m1 2012m1 Bulgaria Hungary Romania World 16

20 Figure 9: Evolution of the unprocessed food HICP index for groups of selected MS - Estonia, Latvia and Lithuania - and of IMF's world agricultural commodity price index Unprocessed food HICP index World price index 2000m1 2002m1 2004m1 2006m1 2008m1 2010m1 2012m1 Estonia Latvia Lithuania World Figure 10: Evolution of the unprocessed food HICP index for groups of selected MS - Croatia, Cyprus, Malta and Slovenia - and of IMF's world agricultural commodity price index Unprocessed food HICP index World price index 2000m1 2002m1 2004m1 2006m1 2008m1 2010m1 2012m1 Croatia Cyprus Malta Slovenia World 17

21 6 RESULTS The Augmented Dickey-Fuller (ADF) and Phillips-Perron (PP) tests statistics were performed for all series used in the analysis - the unprocessed food HICP index, the unemployment rate, the exchange rate, the world agricultural commodity price indices compiled by IMF and ECB and the crude oil price index. The unit root tests were estimated in natural logarithm both in levels and first differences, in order to test for non stationarity in levels of the variables and the order of integration. Most of the series were found to be non-stationary and integrated of order one. The three world agricultural commodity price indices and the crude oil price index are non-stationary and integrated of order one. Country exceptions for the unprocessed food HICP index are Ireland and Romania, which are stationary in levels, for the exchange rate is Romania, which is stationary in levels, and for the unemployment rate are Finland and Germany, which are stationary both in levels and in differences. Engle and Granger cointegration tests were performed between each of the three world agricultural commodity price indices - one compiled by IMF and two compiled by ECB - and the unprocessed food HICP index. Results are summarised in Table 1. Table 1 classifies those MS for which the price series are cointegrated into three categories - MS which have belonged to the euro area from the beginning (plus Greece which acceded in 2001), MS which have recently acceded the euro area and MS which do not belong to the euro area. When considering cointegration tests between the IMF's price index and the unprocessed food HICP index, 22 of the 28 MS exhibit a long run equilibrium relationship between both price series. When considering the ECB's import weighted index, 19 of the 28 MS show a long term relationship between both price series. Finally, when considering the ECB's use weighted index, 17 of the 28 MS show a long term relationship between both price series. Cointegration between both price series seems to be more common among the EU MS when using the IMF's index than when using the ECB's indices. In turn, cointegration between both price series seems to be more common among the EU MS when using the import weighted index than when using the use weighted index. 18

22 Table 1. MS which are cointegrated and non-cointegrated with the three world agricultural price indices according to the cointegration tests Alternative world price indices IMF's WP ECB's importweighted WP ECB's useweighted WP Euro area Cointegrated Austria, Belgium, Finland, France, Germany, Greece, Italy, Luxembourg, Netherlands, Portugal (10/12) Austria, Belgium, France, Germany, Greece, Italy, Luxembourg, Netherlands, Portugal (9/12) Austria, Belgium, France, Germany, Greece, Italy, Luxembourg, Netherlands, Portugal (9/12) Non-cointegrated Spain (1/12) Finland, Spain (2/12) Finland, Spain (2/12) Recently acceded to the euro area Cointegrated Cyprus, Estonia, Latvia, Malta, Slovakia, Slovenia (6/6) Cyprus, Estonia, Malta, Slovakia, Slovenia (5/6) Cyprus, Estonia, Malta, Slovakia, Slovenia (5/6) Non-cointegrated - Latvia (1/6) Latvia (1/6) Cointegrated Bulgaria, Croatia, Denmark, Hungary, Poland, Sweden (6/10) Bulgaria, Denmark, Hungary, Poland, Sweden (5/10) Bulgaria, Hungary, Poland (3/10) Non euro area Non-cointegrated Czech R., Lithuania, UK (3/10) Croatia, Czech R., Lithuania, UK (4/10) Croatia, Czech R., Denmark, Lithuania, Sweden, UK (6/10) Note: WP stands for world agricultural commodity price index. The Euro Area heading refers to MS which have belonged to the Euro Area from the beginning (plus Greece which acceded in 2001). The recently acceded Euro Area heading refers to Slovenia (which acceded in 2007), Cyprus and Malta (which acceded in 2008), Slovakia (which acceded in 2009), Estonia (which acceded in 2011) and Latvia (which acceded in 2014). The non euro area heading refers to Bulgaria, Croatia, Czech Republic, Denmark, Hungary, Lithuania, Poland, Romania, Sweden and UK. Ireland and Romania not included because the series of unprocessed food HICP index are stationary. Table 2 shows the long term price transmission elasticities and the parameters of the error correction terms. As shown, the parameters of the long term price elasticities are positive, as expected. The parameters of the error correction mechanisms are significantly different from zero in most cases and negative, as expected. 19

23 Table 2. Price transmission elasticities for each MS with the three alternative world price indices ECB's import-weighted IMF's WP ECB's use-weighted WP WP MS ECT WPt ECT WPt ECT WPt Austria (0.043) (0.019) (0.044) (0.007) (0.032) (0.013) Belgium (0.037) (0.018) (0.036) (0.014) (0.045) (0.015) Bulgaria (0.033) (0.051) (0.025) (0.039) (0.024) (0.043) Croatia (0.029) (0.029) Cyprus (0.055) (0.025) (0.055) (0.022) (0.036) (0.034) Denmark (0.015) (0.0256) (0.015) Estonia (0.027) (0.037) (0.026) (0.031) (0.020) (0.033) Finland (0.043) (0.027) France (0.048) (0.011) (0.058) (0.01) (0.011) Germany (0.025) (0.023) (0.017) (0.019) Greece (0.046) (0.035) (0.043) (0.024) (0.040) (0.024) Hungary (0.019) (0.046) (0.02) (0.035) (0.017) (0.039) Italy (0.024) (0.018) (0.016) (0.020) Latvia (0.024) (0.044) Luxembourg (0.012) (0.021) (0.015) (0.020) (0.012) (0.023) Malta (0.025) (0.046) (0.031) (0.036) (0.018) (0.039) Netherlands (0.038) (0.023) (0.033) (0.015) (0.016) Poland (0.036) (0.037) (0.024) (0.026) (0.034) (0.032) Portugal (0.039) (0.015) (0.028) (0.018) (0.0314) (0.013) Slovakia (0.048) (0.024) (0.042) (0.019) (0.033) (0.020) Slovenia (0.042) (0.028) (0.041) (0.019) (0.042) (0.019) Sweden (0.043) (0.023) (0.035) (0.015) Note: WP stands for world agricultural commodity price index and ECT for error correction term. Newey West standard errors are shown in parentheses below the parameter estimates. Only countries whose food prices are co-integrated are reported. 20

24 In general, the long term price elasticities are slightly higher when the models are estimated using the IMF's world price index, followed by the ones estimated using the use-weighted ECB's index. The fact that the magnitudes of the elasticities are slightly higher when using the IMF's world price index than when using the other two indices may be due to the commodity coverage of the index - the commodity index compiled by IMF accounts for seafood apart from other commodities whereas the unprocessed food HICP index comprises the fish category. The different commodity coverage may reveal different transmission channels depending on the specific commodity we are considering. On the other hand, the fact that the use weighted index generally yields slightly higher elasticities in magnitude than the import weighted index may be due to the use of a more adapted weighting structure to that of the HICP index. The magnitude of the long term elasticities depend on the MS considered. First-members of the Eurozone present relatively similar long run price transmission elasticities in magnitude and lower than those of the new MS. The lower transmission elasticities in the euro area shows that variations in commodity prices are assimilated to some extent into a reduction in profit margins in the processing and/or the food retail sector. The average size of firms of retail sale of food, beverages and tobacco in non-specialized and specialized stores in the traditional euro area MS are generally bigger than those in the new MS 7, suggesting that the retail sector may be more competitive than that of the new MS (as suggested in Bukeviciute et al 2009). Apart from the fact that the average firm size is larger, in the traditional euro area MS commodity costs take up a smaller proportion of the final product (Bukeviciute et al 2009), which is consistent with having smaller elasticities in magnitude. In evaluating the estimated adjustment parameters, the results show that consumer unprocessed food prices run back to equilibrium slowly after a shock in world commodities markets. Table 3 summarizes the median lag length (in months) for each MS with the alternative World price indices which have been evaluated. The median lag length, which indicates the time it takes for consumer food prices to achieve 50% of the adjustment to their long run equilibrium relationship with world agricultural commodities prices, also show varying effects depending on the MS. Within those MS which present a long term transmission, the lower correspond to France (in a range between 1 and 3 months) and Cyprus and Poland (in a range between 2 and 4 months each). The higher median lag lengths correspond to Italy and Luxembourg. 7 Source: Own calculations with data from EUROSTAT. 21

25 Table 3. Median lag length (months) for consumer food prices to achieve 50% of the adjustment to their long run equilibrium relationship for each MS with the alternative World price indices MS Range of median Lag lengths Note Austria 4-6 Belgium 5-7 Bulgaria 6-8 Croatia 4-5 Only IMF s WP Cyprus 2-4 Denmark 5-6 Only ECB's importweighted WP Estonia 6-12 Finland 4-5 Only IMF s WP France 1-3 Germany Only IMF s WP Greece 3-4 Hungary Italy Latvia n.s. Luxembourg Malta 7-9 Netherlands 4-6 Poland 2-4 Portugal 5-12 Slovakia 2-7 Slovenia 3-5 Sweden 3-4 Only ECB's useweighted WP Note: Only countries whose food prices are co-integrated are reported. 7 CONCLUSIONS The EU is the largest world importer and the second largest exporter of agricultural commodities. Price movements international markets spill over to EU agricultural prices, which are denominated in euro and other floating currencies, and eventually to consumers food prices. Increasing food prices can potentially hurt the most vulnerable EU citizens and households. Since pensions in the EU are becoming increasingly less indexed to inflation, retired citizens and workers paid at standard minimum wages may be more exposed to new food price surges. 22

26 Our work has shown that transmission mechanisms vary across member states, in cointegration significance and levels, long term elasticities and in the speed of adjustment. The differences are striking among MS whose economies are closely integrated, for example Benelux and Germany, Spain and Portugal, and France and Italy. For instance, the time it takes for consumer food prices to achieve 50% of the adjustment to their long run equilibrium relationship with world agricultural commodities prices can last more than 10 months in Germany, Italy and Luxembourg, whereas in France, The Netherlands and Belgium the length is six months or less. Furthermore the transmission elasticities also vary significantly across countries. It tends to be below 0.25 in the Eurozone, and above 0.5 in the nonmembers of Eurozone, suggesting that monetary policy and exchange rate stability provide a stronger cushion for world food price instability. But differences between MS of the Eurozone attest for significant differences of eating and purchasing habits, which are cause and effect of significant structures of their food retail sectors and food industry competitions. More work is needed to investigate further the relevance and representativeness of the consumer food prices for the actual cost of food to consumers. Recent literature suggest that consumer prices vary more than the index suggests, and the consumption basket is also adjusted in the event of higher food prices. More micro-data analyses are needed in order to evaluate the consequences of agricultural markets volatility in consumers, specially on the poorest households. Additional variables should be introduced in the price transmission analysis, including those referred to competitive market structures in the food sector which might give insight to a deeper understanding of the differences in price transmission among EU MS. 23

27 References Apergis, N., Rezitis, A. (2011). Food Price Volatility and Macroeconomic Factors: Evidence from GARCH and GARCH-X Estimates. Journal of Agricultural and Applied Economics, 43(1): Baquedano F.G., Liefert W.M., Shapouri S. (2011). World market integration for export and food crops in developing countries: a case study for Mali and Nicaragua. Agricultural Economics 42: Baquedano F.G., Liefert W.M. (2014). Market integration and price transmission in consumer markets of developing countries. Food Policy 44: Berck P., Leibtag E., Solis A., Villas-Boas, S. (2009). Patterns of Pass-through of Commodity Price Shocks to Retail Prices. American Journal of Agricultural Economics 91(5): Bukeviciute, L., Dierx, A., Ilzkovitz, F. (2009). The functioning of the food supply chain and its effect on food prices in the European Union. European Economy Occasional Papers 47. Busicchia, B. (2013). The Question of a Reasonable Price for Food: Policy Alternatives to Control Food Price Inflation in Developed Economies. In: Farmar-Bowers, Q., Higgins, V., Miller, J. (Eds), Food Security in Australia: Challenges and prospects for the future, Springer. Davidson, J., Halunga, A., Lloyd, T.A., McCorriston, S., Morgan, C.W. (2011). Explaining UK Food Price Inflation in TRANSFOP Working Paper No.1, TRANSFOP project, EU 7th Framework Programme, Grant Agreement No. KBBE TRANSFOP, De Boef, S., Keele, L. (2008). Taking time seriously. American Journal of Political Science 52(1): Engle, R.F., Granger, C.W.J. (1987). Co-Integration and Error Correction: Representation, Estimation and Testing. Econometrica 55: European Central Bank (2011). Structural features of distributive trades and their impact on prices in the Euro Area, Occasional Paper Series No 128. European Commission (2008). Subject: Update on recent price developments in EU-27 agriculture and food retail. Note for the file, Brussels. 24