Market Imperfections in the European Food Processing Industry

Transcription

1 Internatonal comparson of product supply chans n the agr-food sector: determnants of ther compettveness and performance on EU and nternatonal markets WORKING PAPER N 12, September 2014 Market Imperfectons n the European Food Processng Industry Authors: Lukáš Čechura 1, Henrch Hockmann 2, Zdeňka Kroupová 1 1 Czech Unversty of Lfe Scences, Prague, Czech Republc 2 Lebnz Insttute for Agrcultural Development n Transton Economes, Halle (Saale), Germany Abstract The paper analyzes the degree of market mperfectons n the nput and output food processng markets usng the derved mark-down and mark-up models and employng stochastc fronter methodology. The research was based on data drawn from the Amadeus database. The estmated mark-down model revealed some degree of non-compettve behavour n the nput food processng market for all analysed sectors (slaughterng, frut and vegetables, dary and mllng) wth dfferent degree of market mperfectons among the sectors. The EU slaughterng common market s charactersed by sgnfcantly greater market mperfectons as compared to the dary and mllng sectors. The results of the ftted mark-up model suggest that market mperfectons on the output market are not so pronounced for the slaughterng sector. The degree of market mperfectons s hgher for the output market n the dary and mllng sectors as compared to the nput market. The fruts and vegetables sector has almost the same level of market mperfectons for the output as for the nput market. COMPETE Workng Papers present work beng conducted wthn the COMPETE research project. The project analyses compettveness of varous European agr-food supply chans and ts determnants. Workng papers are ntended to stmulate reactons from other experts n the feld. For more nformaton on the project see the back cover. Unless otherwse ndcated, the vews expressed are attrbutable only to the authors n a personal capacty and not to any nsttuton wth whch they are assocated. Avalable for free downloadng from the COMPETE webste ( Copyrght 2014, Čechura, Hockmann, Kroupová COMPETE Project Coordnaton: Lebnz Insttute of Agrcultural Development n Transton Economes (IAMO) Theodor-Leser-Str.2 D Halle Telephone: Fax: Emal:compete@amo.de

2 Contents 1 Introducton Theoretcal framework and estmaton strategy Theoretcal framework Mark-down model Mark-up model Statc vs. dynamc settng Estmaton strategy Data Results Mark-down model Mark-up model Conclusons References COMPETE WORKING PAPER 2

3 Tables Table 3.1 Structure of the data set Table 4.1 Parameter estmates Table 4.2 Relatve mark-down - Slaughterng Table 4.3 Relatve mark-down Fruts and vegetables Table 4.4 Relatve mark-down Dary Table 4.5 Relatve mark-down Mllng Table 4.6 Parameter estmates Table 4.7 Relatve mark-up Slaughterng Table 4.8 Relatve mark-up Fruts and vegetables Table 4.9 Relatve mark-up Dary Table 4.10 Relatve mark-up Mllng COMPETE WORKING PAPER 3

4 Market Imperfectons n the European Food Processng Industry Lukáš Čechura, Henrch Hockmann, Zdeňka Kroupová 1 Introducton The food processng ndustry, an mportant part of the agr-food chan, creates the demand for food n general, and agrcultural raw materals n partcular, and has the potental to substtute the ncentves mplemented by agrcultural polcy for ncentves resultng from the demand for food. In ths respect, at least two requrements have to be met so that the processors can ncur ths functon. Frst, market power must not be a crucal factor n the exchange transactons of raw materals, so that processors are not able to extract rents from farmers. Second, food processors must be hghly productve, and thus compettve, so that they can create a hgh demand whch can be met by the domestc agrcultural sector. In ths workng paper, we focus on the frst requrement and analyse the market mperfectons n the EU food processng ndustry. Our am s to conduct a comparatve analyss among the dfferent EU countres and dfferent ndustres, and dentfy the degree of market mperfectons n both the nput and output food processng markets. That s, based on the derved mark-down and mark-up model and usng stochastc fronter methodology, we wll dentfy the degree of non-compettve behavour n the slaughterng, fruts and vegetables, dary and mllng sectors n 24 EU member states (only Croata, Cyprus, Luxemburg and Malta are mssng). Ths study extends the research on the analyss of market mperfectons. Snce 1980 there have been numerous studes based on a New Emprcal Industral Organzaton theoretcal background focused on detectng market power or, n general, market mperfectons n the agrcultural and food processng market, as the case may be. Most of these studes are based on market-level data. There s evdence of the olgopsonstc market power of food processors (e.g. Schroeter and Azzam, 1990, Morrson Paul, 2000), but there are also studes whch faled to fnd any evdence of olgopsonstc power (e.g. Welwta and Azzam, 1996, Muth and Wohlgenant, 1999, Perekhozhuk and Grngs, 2006) or found only weak olgopsonstc power (Scalco and Brage, 2014). Moreover, McCorrston (2002) concluded that food markets are now more typcally olgopolstc. Studes usng frm-level data are not so numerous and nclude, for example, Hockmann and Vönek (2009), Bakucs et al. (2009), Perekhozhuk et al. (2011) and Acharya et al. (2011). In ths workng paper, we use frm-level data to complement the research on market mperfectons by dentfyng the degree of non-compettve behavour n the food processng ndustres of the 24 EU member states. In partcular, the paper addresses the followng research questons: COMPETE WORKING PAPER 4

5 1. The frst queston relates to market mperfectons n the nput food processng market. The am s to dentfy the degree of non-compettve behavour of the food processors wth respect to farmers. 2. The second queston concerns market mperfectons n the output food processng market. The am s to dentfy the degree of olgopolstc market power. 3. The last queston relates to the country specfcs, especally whether the nput and output processng markets dffer sgnfcantly among countres, and whether the EU processng market s becomng ncreasngly compettve or whether an dosyncratc development of market power can be observed. The workng paper s organzed as follows: Chapter 2 contans the theoretcal framework and presents the estmaton strategy; Chapter 3 descrbes the data set; Chapter 4 presents the results of the mark-down and mark-up model, compares the estmated market mperfectons and ther development, and dscusses ther causes and consequences. Chapter 5 contans concludng remarks and polcy recommendatons. COMPETE WORKING PAPER 5

6 2 Theoretcal framework and estmaton strategy The research questons wll be dealt wth (1) estmaton of country specfc nput dstance functon wth nter- and ntra-sectoral heterogenety usng the Amadeus database and (2) Based on the parameters the effcent output level wll be calculated. These wll be used n a metafronter estmate of sectoral nput dstance functons to determne the TFP level of the analysed sectors (slaughterng, fruts and vegetables, dary and mllng) and TFP development. In order to produce coherent results, all models (the country-specfc models n (1) as well as the metaproducton models n (2) wll make use of the same procedure: The models are formulated as a specal case of nput dstance functons wth sector dummes n (1) for capturng the ntersectoral dfferences n technology. Moreover, n all models t s explctly consdered that food processng producton possbltes are affected by frm (ntrasectoral) heterogenety, whch nfluences the level as well as shape of the producton possbltes. 2.1 Theoretcal framework The research questons wll be addressed by estmatng the derved mark-down and mark-up model and employng stochastc fronter methodology. The stochastc fronter approach for detecton of the degree of monopoly power was frst appled by Kumbhakar et al. (2012). The novelty of our study s the dervaton of a mark-down model usng stochastc fronter methodology for detecton of the abuse of olgopsonstc/monopsonstc behavour. Moreover, we extend the theoretcal backgrounds of the mark-up model and dscuss the statc versus dynamc (game theory) nterpretaton of results Mark-down model The mark-down model s derved usng the conjectural varaton approach. Frst, we ntroduce the frm optmzaton problem, whch s followed by the dentfcaton and estmaton of the mark-down. A) Frm optmzaton We follow the methodology developed by Bresnahan (1982 and 1989) and Muth, Wohlgenant (1999) to test for olgopsonstc market power. The proft of a processor () s gven by: (1) R( p, x, z, t) w x wz ' z X where p s a vector of product prces, R(p, x, z, t) represents the revenue functon dependng n addton on the agrcultural raw materals (x), other nputs (z) and a tme trend (t) as an ndcator of techncal change. The symbol w s used for the correspondng factor prces. The supply functon of raw materals s: (2) x g(, s) or g 1 ( x, s) w x w x Here, s s a vector of supply shfters and x s the total supply of raw materal. However, for analysng the optmal demand of the processor t s more convenent to use the nverse supply functon w x g 1 ( x, s). Gven (1) and (2), the frst order condton for proft maxmsaton s: COMPETE WORKING PAPER 6

7 1 R( p, x, z) g ( x, s) x (3) wx x 0, x x x where n processor 's demand for mlk. After rearrangement of (3): (4) w X x / x represents the ncrease n total farm supply nduced by an ncrease R( p, x, z, t) 1, x 1 x g ( x, s) ln x where x 0 denotes the prce elastcty of the raw mlk supply 1 g ( x, s) x ln wx x x and s an elastcty capturng the degree of olgopsonstc market power x x (Bresnahan, 1989). The parameter range s 0 < <1. 0 corresponds to perfect competton, whle 1 characterzes a monopsonstc market. 1 From (4) t follows that: w x R MRPx x Ths relatonshp can furthermore be expressed as: (5) w x x x R x ln R ln MRP D x R R x R ln x ln x o Where the last equalty comes from the dualty of the revenue (R) and output dstance (D o ) functons. B) Estmaton and dentfcaton of the mark-down Inequalty n (5) can be transformed nto equalty by addng a non-negatve one-sded term, u: (6) o wx x ln D u R ln x, u 0. Assumng that the output dstance functon has a translog form: (7) 1 1 D ß ß t ß t ß x ß xt ß x ßz ' ln z ßzt ' ln zt ln z ' Bzz ln z ln z ' ßzx ln x 2 1 ßy ' ln y ßyt ' ln yt ln y ' Byy ln y ln y ' ßyx ln x 2 ln y ' B ln z o 2 ln 0 t tt x ln xt ln xx ln yz 1 Snce prces of other nputs are assumed to be constant, ther optmal level s gven when the factor prce s equal to the value of margnal revenue: w Z R( p, x, z, t) z. COMPETE WORKING PAPER 7

8 Wth the correspondng dfferental and addton of statstcal nose (v), (6) becomes: wx x (8) ßx ßxtt ßxx ln x zx 'ln x 'ln u v R ß z ß y y. For one output, (8) reduces to (homogenety of degree 1 requres that yx = 0): wx x (9) ßx ßxtt ßxx ln x zx 'ln u v R ß z. Snce we defne the relatve mark-down by: (10) MRPx -w x MRP x It can be estmated va (expandng by x/r and usng the dualty relatonshp): (11) That s, (11 ) u o ln D / ln x u ß ß t ß ln x ß 'ln z. x xt xx zx Mark-up model The mark-up model can be derved n the same manner as the mark-down model. A) Frm optmzaton The optmzaton problem for the output market s analogous to the nput market. In ths case, the proft functon of processor () s gven by: (12) p y C( w, y, t) where p s a prce of output, y s the output of processor (), w s a vector of nput prces, and C(w,y,t) s a cost functon of processor () and tme trend (t) for capturng techncal change. The demand or nverse demand functon, respectvely, s: 1 (13) y f ( p, d) or p f ( y, d) where d s a vector of demand shfters and y s the total demand for food. Gven (12) and (13), the frst-order condton for proft maxmsaton s: 1 f ( y, d) y C( w, y, t) (14) y p 0 y y y COMPETE WORKING PAPER 8

9 Relaton (14) can be rearranged as: (15) C y t p ( w,, ) 1 p y, y p y y where p 0 s a demand elastcty of the fnal product and 1 f ( y, d) y y y s a conjectural elastcty capturng the degree of olgopolstc market power. The parameter s n the nterval 0;1. 0 ndcates compettve behavour and 1 characterzes monopolstc power. That s, a postve value of (0;1 ) ndcates the presence of noncompettve behavour n the output market. In partcular, the hgher s, the greater s the degree of olgopolstc market power, or the degree of non-compettve behavour n general. It follows from (15) that: C(w, y, t) p y for 0;1. Ths relaton can be expressed as: (16) p y C C( w, y, t) y y C ln C ln y ln D ln y I, where the last equalty comes from the dualty of the cost (C) and nput dstance (D I ) functons. B) Estmaton and dentfcaton of the mark-down Inequalty n (16) can be transformed nto equalty by addng a non-negatve one-sded term, u: p y C ln D ln y (17) u I, u 0. Assumng that the nput dstance functon has a translog form: (18) ln D I tt ttt y ln y yt ln yt yy (ln y) α x ln ~ x α xt ln ~ xt ln ~ x A xx ln ~ x ln ~ x A xy ln y 2 where x~ j x / x for j = 1,,J. j J COMPETE WORKING PAPER 9

10 Then ln D ln y I t ln y α ln ~ x u v. y yt yy xy Consequently, t follows from (17) and (18) that the functon to be estmated has the form: p y (19) y ytt yy ln y α xy ln ~ x u v. C Snce we defne the relatve mark-up as: (20) p MC. MC It can be estmated va: (21) u I ln D / ln y That s, (21 ) uˆ ˆ t ln y α lnx. y yt yy yx 2.2 Statc vs. dynamc settng 2 Conjectural elastctes derved n the mark-down and mark-up models n the precedng secton mght be msleadng as an ndcator of market power. Moreover, ther theoretcal consstency s rather poor. In ths secton, a dfferent approach wll be presented. No functonal relatonshp between an ndcator and market power wll be derved, but the varous factors facltatng and hamperng colluson n the food processng ndustry wll be dscussed. That s, the dfferent reasons or nterpretatons for the dentfed noncompettve behavour or market mperfectons usng the derved mark-down (8) and mark-up (19) models wll be presented, respectvely. Frm behavour n a dynamc settng: Repeated games Whle theory provdes unambguous results regardng a frm s behavour n a statc settng (e.g. Cournot equlbrum for full capacty utlsaton, Bertrand equlbrum for underutlsed capactes), a frm s behavour n a dynamc framework s far from unambguous. Game theory results exst for fntely and nfntely repeated games. In the frst case, backward nducton suggests that frms behave as n a smultaneous-shot game. The soluton becomes more complex when the game s repeated ndefntely. 2 In ths secton, we focus our attenton on the output market snce t s usually found n the lterature. However, the correspondng nterpretaton for the nput market s straghtforward. COMPETE WORKING PAPER 10

11 Accordng to the Folk theorem, every outcome that s better than the worst Nash equlbrum can be reached by an equlbrum strategy (Fgure 1). Ths mples that there exst plenty of equlbrum solutons, and no defnte answer regardng the behavour of a frm n the ndustry s possble (Fudenberg, Trole, 1990). It may be argued that the assumpton of an ndefntely repeated game s artfcal; however, ths requrement may be weakened by the assumpton that the probablty of contnuaton from one stage to the next s strctly postve. Fgure 1: Equlbrum soluton n a duopoly under dfferent behavoural assumptons 2 2 monopoly Possble outcomes wth perfect colluson 2 Cournot Possble Nash equlbra n nfntely repeated games 2 =0 (Bertrand) 1 =0 (Bertrand) 1 Cournot 1 1 monopoly Source: Trole (1988) Factors hnderng and facltatng colluson Havng defned the prncple soluton set for a stuaton wth dynamc nteractons, t remans to dscuss the determnants that may hnder and facltate a soluton. Possble nfluences on colluson wll be dscussed wth an llustratve model. We assume a homogeneous good ndustry wth n frms. Furthermore, demand growth wth rate g. 3 When frms coordnate ther prcng behavour, prces wll be larger than margnal costs (p>mc) and ndustry profts C wll be realzed, whch are dvded by a predefned key that conssts of the proft shares receved by the ndvdual frms (s ). 4 A frm devatng from ths (mplct) agreement sets a prce slghtly lower than p and receves profts C n the devatng perod and C thereafter, wth >. In addton, we assume that > s > snce otherwse there would be no ncentve to devate (the frst nequalty) and no ncentve to cooperate (the second nequalty). The parameters I and I reflect the producton capactes of the frms. The larger I,.e. the less the capacty of the devatng frm s constraned, the easer the frm can serve the whole market. By contrast, hgher I ndcate large capacty constrants of the compettors. The dscount factor s gven by. The frequency wth whch transactons occur s gven by. The hgher the, the hgher the frequency,.e. = 1 corresponds to an annual transacton whle = 365 ndcates daly transactons. 3 4 We follow the procedure usually found n the lterature, where colluson s dscussed wth regard to output markets. However, a correspondng nterpretaton for nput or procurement markets s straghtforward. The s can be thought of as the ntal market shares or shares of the producton capactes of the frms. COMPETE WORKING PAPER 11

12 Next, we wll apply the followng strategy. Frms agree to cooperate n the frst stage. As long as there s no devaton from the agreed process, cooperaton wll contnue. However, f one frm devates, the agreement wll break down and frms wll begn to apply compettve prcng. Formally, ths strategy s sustanable, when the followng condton holds: j C C C (22) s 1 g 1 g, j0 j1 Thus, colluson s a vable strategy when frms put more weght on future than on present profts. Moreover, (22) can be transformed to present a threshold at whch colluson s sustanable: (23) s g a *, 1 Generally, colluson s easer to sustan when the threshold s lower, because n that case even an mpatent frm wth a low dscount factor regards colluson as a benefcal strategy. Condton (23) provdes that the threshold s ncreasng n and, but decreasng n s, g and. The potental for colluson n the food ndustry If the frequency of transactons () s hgh t makes colluson more lkely (Bugues and Rey, 2004, p. 93). Moreover, ths stuaton mples that market transparency s also relatvely hgh (Ivald et al., 2003, p. 22). Processors can detect possble devaton from an agreement relatvely easly from a change n delveres or negotatons wth farmers. In addton, prce nformaton systems are avalable that provde nformaton about recent prce developments wthout serous delays. These suggest that devaton may be detected mmedately, and the compettors can react quckly. Thus a hgh degree of market transparency reduces the ncentves to devate from a collusve agreement. A further characterstc of food producton s the pershableness of the raw materal. Raw materal usually cannot be stored for a long tme wthout beng processed. Ths puts farmers n a poor barganng poston, because the opportuntes to adjust producton mmedately to changng market condtons are rather lmted. From ths t follows that processors possess dstnct possbltes of ganng from opportunstc behavour,.e. of extractng large parts of the producer rents assocated wth producton (Ivald et al., 2003, p. 50). Moreover, condton (23) provdes that market growth facltates collusve behavour, because devaton would lead to hgher foregone future profts. In general, the costs of capacty underutlsaton determne a frm s behavour. An ncrease n the costs of capacty underutlsaton ncreases the ncentves to swtch to compettve behavour over tme. Furthermore, the ncentves for large and small frms to devate from colluson, as well as the consequences of devaton, wll vary by frm sze (Compte, et al., 2002). In condton (23), these forces fnd ther expresson n s, and. Small frms wll have lttle s and. On the one hand, the frst reduces whle the second ncreases the ncentves to collude. However, t can be expected that, due to capacty constrants, s not much larger than s. Correspondng to condton (23), the threshold wll be relatvely low,.e. together the two parameters facltate colluson. In addton, f a small frm devates, the larger compettors may j COMPETE WORKING PAPER 12

13 react wth severe competton. Ths would result n an ncrease n the procurement prce, whch n turn because of the low economes of scale n small frms suggests a rather low or even a market ext ( = 0). Summarzng these arguments suggests that small frms may have few ncentves to devate from a collusve agreement. On the other hand, even f the frms devate, the sanctons may be less severe than those whch occur when a large frm or frms devate. Snce small frms usually have low capactes, ther addtonal demand when devatng wll be low, possbly wthout a notceable mpact on market prces. Ths argument holds as long as not too many small companes try to deceve. In large frms where colluson may lead to hgh unused capactes, can be expected to be sgnfcant larger than s, whch mples a hgh threshold. However, these low ncentves to collude can be compensated by the threat of ntense competton on the raw materals market (low ), whch n turn requres that large frms possess suffcent underutlsed capactes to make the threat credble. Ths dscusson of the varous determnants of colluson shows that there s no unambguous answer to the queston of whether the stuaton n the food processng ndustry facltates or hnders colluson. However, n our vew, the ponts n favour of colluson (hgh degree of market transparency, hgh frequency of nteracton, small number of large frms whch could actually nfluence market prces, threat of severe sanctons due to low capacty utlsaton, opportunstc behavour) make collusve behavour more lkely than compettve behavour Estmaton strategy The derved mark-down model (8) and mark-up model (19) wll be estmated usng the stochastc fronter methodology. Snce we respect both the heterogenety n producton structures and possble tme-varyng mark-down or mark-up component, we employ the Fxed Management model (Álvares et al., 2003 and 2004). Fxed Management model Álvarez et al. (2003 and 2004) specfed the Fxed Management model as a specal case of the Random Parameters model n the followng form: Mark-down model (for one output): (24) wth u t ln f wx x x xtt R * 1 mm 2 xx mm ln x m *2 t β tm zx ln z * t m t x, z, t, m ; β ln f x, z, t, m ; β 0 t t t t xm m * x t β zm m ln z * t u t v t 5 Snce the dataset consst of large companes (see chapter 3) n majorty of cases we can expect that some large companes may exercse market power. COMPETE WORKING PAPER 13

14 Mark-up model: (25) p y C t y m m * yt 1 2 yy mm ln y m *2 t α tm xy ln ~ x * t m t ym m * y t α xm m ln ~ x * t u t v t. wth y, ~ x, t, m ; α ln gy, ~ x, t, m ; α 0 u ln g, t t t t t The mark-down or mark-up component, u t 0, captures the devatons from compettve behavour. m ~ 0,1 represents unobservable fxed management. The symbol * expresses that m could possess any dstrbuton wth zero mean and unt varance. The dfference between real (m ) and optmal ( m ) management determnes the degree of market mperfectons exercsed by the -th producer. Álvarez et al. (2004) showed that u t can be estmated accordng to Jondrow at al. (1982) as (26), wth smulated m * accordng to (27). (26) Eu m u 2 where, (27) v t t t m / t m,, 2 1 t m / 2 2 and t t t u v v u. R 1 ˆ wx, x * m, r f ln xt, z t, t, m, r ; β ˆ wx, x R r 1 R E m, x, z, β or R R 1 ˆ wx, x * f ln x, z,, ; β t t t m R r1 R R 1 p y, ~ * m ˆ ln x,, ; α, ~, r g yt t t m, r ˆ p y R r 1 C E m, y x, α, respectvely. R C 1 p y ˆ ln, ~ * g y x,, ; α t t t m R r1 C The Fxed Management model s ftted by maxmum smulated lkelhood wth NLOGIT 5.0. COMPETE WORKING PAPER 14

15 3 Data The data we use n the analyss s drawn from the Amadeus database, created and produced by Bureau van Djk. The database contans fnancal nformaton for publc and prvate companes across Europe. 6 The database provdes detaled nformaton about (standardsed) annual accounts, fnancal ratos, sectoral actvtes and ownershp nformaton. 7 Table 3.1 Structure of the data set EU member country Slaughterng Frut & Vegetable Dary Mllng Austra Belgum Bulgara Czech Republc Germany Denmark Estona Span Fnland France Unted Kngdom Greece Hungary Ireland Italy Lthuana Latva Netherlands Poland Portugal Romana Sweden Slovena Slovaka Source: Amadeus database and our own calculatons The panel data set that we use n our analyss contans companes whose man actvty s food processng accordng to the NACE classfcaton (NACE 10 manufacture of food products groups from 10.1 to 10.9). It s an unbalanced panel data set whch represents the perod from 2003 to 2012 and contans 9,885 food processng companes from 24 EU countres. Snce not all companes n the database have complete nformaton, we exclude those companes wth negatve and zero values of the varables of nterest. Thus, we were 6 The dataset conssts of the companes who are oblged to publsh a balance sheet and a proft loss account (cooperatves, jont stock companes, etc.). That s, the dataset contans manly large (and often successful) companes that mght be able to exercse market power (see chapter 2). In other words, the sample of companes s based wth respect to the companes that mght be able to explot ther barganng power. 7 More nformaton on the Amadeus database s provded at: COMPETE WORKING PAPER 15

16 constraned to usng an unbalanced panel data set contanng 8,110 companes wth 52,682 observatons coverng the perod from 2003 to Moreover, we concentrated on the analyss of four food processng sectors: slaughterng (12,239 observatons), fruts and vegetables (5,541 observatons), dary (6,367 observatons) and mllng (3,271 observatons). Table 1 presents the structure of the data set. The followng varables were used n the analyss: Mark-down model: Cost share = Materal costs/revenue, Materal, Captal and Labour. Materal costs were used n the form of the total costs of materals and energy consumpton per company. Revenue s represented by operatng revenue (Turnover) of the company. Materal s the total costs of materals and energy deflated by the ndex of producer prces n the ndustry (country level; 2010 = 100). Labour s represented by the total number of employees and Captal s the book value of fxed assets deflated by the ndex of producer prces n the ndustry (country level; 2010 = 100). Mark-up model: Revenue share = Revenue/Costs, Output, normalzed Materal and Labour. Revenue s represented by operatng revenue (Turnover) of the company. Costs are the sum of Labour costs, Materal costs and Captal costs. Labour costs are represented by the costs of employees, Materal costs are the total costs of materals and energy consumpton per company, and Captal costs are calculated as the book value of fxed assets multpled by the nterest rate accordng to convergence crtera. Output s represented by operatng revenue (Turnover) of the company and s deflated by the sectoral ndex of food processng prces (EU level 27 countres or country level f t was dsposable, respectvely; 2010 = 100). Materal and Labour are normalzed by Captal. As n the case of the markdown model, Materal s the total costs of materals and energy deflated by the ndex of producer prces n the ndustry (country level; 2010 = 100). Labour s represented by the total number of employees and Captal s the book value of fxed assets deflated by the ndex of producer prces n the ndustry (country level; 2010 = 100). Moreover, we rejected producers wth fewer than three observatons (on average) to decrease the problem assocated wth the entry and ext of producers from the database. COMPETE WORKING PAPER 16

17 4 Results 4.1 Mark-down model Table 4.1 provdes the parameter estmate of the mark-down model for the slaughterng, fruts and vegetables, dary and mllng sectors n 24 EU member states (only Croata, Cyprus, Luxembourg and Malta are mssng). Snce we assume both heterogenety n technology and tme-varyng market mperfectons, we employ the Fxed Management model, whch allows us to capture unobservable heterogenety and has a tme-varyng, onesded error component (see Chapter 2). We estmated models separately for the slaughterng, fruts and vegetables, dary and mllng sectors. As could be expected, all the ftted parameters are hghly sgnfcant, n the majorty of cases even at a 1 % sgnfcance level. Ths holds for all sectors. Moreover, the hgh sgnfcance of the coeffcents on unobservable fxed management suggests that the chosen specfcaton approxmates well the estmated relatonshp, and that heterogenety among frms s an mportant characterstc of the food processng sector. The ftted parameters show that Captal and Labour have a negatve mpact on the materal cost share. Ths also holds for the tme varable n the slaughterng and dary sectors. That s, the cost share decreases over tme n the slaughterng and dary sectors and ncreases n the mllng sector. The tme varable s not sgnfcant n the fruts and vegetables sector. The negatve mpact of captal and labour nputs on the materal cost share and, on the other hand, the postve contrbuton of the materal nputs mples that the larger companes produce wth smaller relatve value added. The unobservable heterogenety component (management) contrbutes postvely to the cost share n slaughterng and dary and negatvely n the fruts and vegetables and mllng sectors. Moreover, the postve mpact s deceleratng, and the negatve accelerates over tme. The ncrease n management has a dfferent effect n each sector. Slaughterng s characterzed by the postve contrbuton of management on the mpact of captal, as well as labour on the materal cost share. The opposte holds true for the tme component and materal nputs. The heterogenety component negatvely determnes the mpact of captal on the materal cost share, whereas t postvely determnes the labour and materal mpact n fruts and vegetables. In the dary sector, the ncrease n the heterogenety component contrbutes postvely to the mpact of captal and labour on the materal cost share and negatvely to the mpact of materal. Fnally, the mllng sector s characterzed by a negatve contrbuton of the heterogenety component to the mpact of labour nputs on the materal cost share, and a postve contrbuton on the mpact of captal and materal. In terms of the relatonshp between the heterogenety component and mark-down component or, n general, market mperfectons, as the case may be, an ncrease n the heterogenety component leads to an ncrease n the mark-down component or market mperfectons n the fruts and vegetables and mllng sectors and a decrease n slaughterng and dary, respectvely. Moreover, hgher captal leads to an ncrease n the mark-down component n the slaughterng, dary and mllng sectors for a gven level of management. The opposte holds true for fruts and vegetables. Hgher labour determnes postvely the mark-down n slaughterng, fruts and vegetables and dary, and negatvely n the mllng sector. Fnally, the hgher the materal nputs, the smaller the market mperfectons COMPETE WORKING PAPER 17

18 for a gven level of management n slaughterng and dary, and the hgher they are n the fruts and vegetables and mllng sectors. Table 4.1 Parameter estmates Slaughterng Means for random parameters Coeffcent on unobservable fxed management Varable Coeff. SE P [ z >Z*] Varable Coeff. SE P [ z >Z*] Const Alpha_m Tme Tme Captal Captal Labour Labour Materal Materal Alpha_mm Sgma Lambda Fruts and vegetables Means for random parameters Coeffcent on unobservable fxed management Varable Coeff. SE P [ z >Z*] Varable Coeff. SE P [ z >Z*] Const Alpha_m Tme Tme Captal Captal Labour Labour Materal Materal Alpha_mm Sgma Lambda Dary Means for random parameters Coeffcent on unobservable fxed management Varable Coeff. SE P [ z >Z*] Varable Coeff. SE P [ z >Z*] Const Alpha_m Tme Tme Captal Captal Labour Labour Materal Materal Alpha_mm Sgma Lambda Mllng Means for random parameters Coeffcent on unobservable fxed management Varable Coeff. SE P [ z >Z*] Varable Coeff. SE P [ z >Z*] Const Alpha_m Tme Tme Captal Captal Labour Labour Materal Materal Alpha_mm Sgma Lambda Source: own calculatons COMPETE WORKING PAPER 18

19 These estmate correspond to our expectatons only partally. The heterogenety (management) component can be vewed as a measure of the qualty of nputs as well as a measure of good practses. That s, hgher qualty and better practses (management, strategy) can be a source of non-compettve behavour. Moreover, hgher nputs are assocated wth the larger sze of a food processor, and wth larger sze (or market share n general) we should expect a hgher relatve mark-down ether due to the hgher market power or, n terms of game theory, the hgher probablty of colluson. The frst assumpton s met n the fruts and vegetables and mllng sectors. That s, the hgher heterogenety component s a source for hgher non-compettve behavour n these sectors. Moreover, the sgns of unobservable fxed management on nputs are also only partally n lne wth the second assumpton for a gven level of management. But f we regress the relatve mark-down on the nput quanttes we can fnd a sgnfcant postve relatonshp. To sum up, the hypothess about the postve relatonshp between market power or market mperfectons and sze cannot be rejected; however, t s sgnfcantly determned by the sources of non-compettve behavour (see the dscusson n Chapter 5). The mpact of tme on the relatve mark-down s negatve n the slaughterng, fruts and vegetables and dary sectors and postve n the mllng sector. Ths could be a sgn of ncreasng compettveness on the EU common market n the slaughterng, fruts and vegetables and dary sectors. Fnally, the parameter λ s hghly sgnfcant and s greater than one n the slaughterng, dary and mllng sectors, and approxmately one n fruts and vegetables. Ths means that the varaton n the mark-down component u t s more pronounced than the varaton n the random component v t n slaughterng, dary and mllng. That s, the estmates ndcate that dfferences n non-compettve behavour among food processors are mportant characterstcs of these sectors. Table 4.2 provdes statstcal characterstcs of the relatve mark-down for slaughterng. The relatve mark-down s n the nterval zero to one. Zero ndcates no market mperfectons or generally compettve behavour, as the case may be,.e. the stuaton where margnal revenue product equals the prce of the materal nputs (especally agrcultural raw materal, whch domnates the materal nputs n the analysed food processng sectors). Then, the postve value of the relatve mark-down represents non-compettve behavour. In partcular, an ncreasng relatve mark-down s assocated wth ncreasng market mperfectons or, n general, ncreasng abuse of market power,.e. the food processor has a greater degree of olgopsonstc power (e.g. due to the hgher barganng power) to charge mark-down (MRP x > P x ) wth respect to supplers (n ths case farmers). Another nterpretaton of the MRP x > P x s n terms of game theory,.e. coordnaton of the frms prcng behavour colluson. Wth respect to the dfferent nterpretaton of the surplus of margnal revenue product over the nput prce, we wll relate the ncrease n relatve markdown to an ncrease n the degree of non-compettve behavour, whch s more general compared to the ncrease n olgopsonstc power nterpretaton. 8 8 The nterpretaton n terms of olgopsonstc power can be msleadng; see the dscusson n Chapters 2.2 and 5. COMPETE WORKING PAPER 19

20 The estmated overall mean of the relatve mark-down for the EU slaughterng common market, , ndcates non-compettve behavour n the EU slaughterng ndustry. The dstrbuton of the relatve mark-down s relatvely narrow, wth standard devaton of 0.11, and also slghtly skewed toward smaller values. The frst decle ndcates that 10 % of producers have a rather low mark-down. On the other hand, the last 10 % of producers reach a relatve mark-down hgher than 0.28, ndcatng a consderably large relatve markdown and thus a degree of non-compettve behavour. Table 4.2 Relatve mark-down - Slaughterng Country Statstcal characterstcs of relatve mark-down Mean Std.Dev Mn. Max. 1 st Decle 9 th Decle 1 st Quartle 3 rd Quartle Cases Total Austra Belgum Bulgara Czech Republc Germany Denmark Estona Span Fnland France Unted Kngdom Greece Hungary Italy Ireland Latva Lthuana Netherlands Poland Portugal Romana Sweden Slovena Slovaka Source: own calculaton The overall means of the relatve mark-down dffer among the ndvdual member states, and the dfferences are qute large. The means of the relatve mark-down can be found n the nterval 0.05 to Bulgara (0.14), the Czech Republc (0.11), Denmark (0.12), COMPETE WORKING PAPER 20

21 Estona (0.11), Unted Kngdom (0.10), Greece (0.09), Ireland (0.09), Latva (0.05), Lthuana (0.06), Poland (0.09), Portugal (0.13), Romana (0.14), Slovena (0.13) and Slovaka (0.09) have a lower mean compared to the EU average. That s, we can fnd smaller market mperfectons n these countres. In partcular, farmers may face a lower degree of noncompettve behavour n these countres as compared to Austra (0.22), Belgum (0.19), Germany (0.23), Fnland (0.27), France (0.21) and Italy (0.18), whch are the countres wth a mean of relatve mark-down hgher than the EU average. The dstrbuton of the relatve mark-down s narrow n all countres and slghtly skewed toward small values. That s, despte the small standard devatons, sgnfcant dfferences between the 1 st and 9 th decles are pronounced n the majorty of EU countres. Large dfferences can be found prmarly n Austra, Belgum, Bulgara, Germany, Span, Fnland, France, Hungary, Italy, the Netherlands, Romana and Sweden. Whereas the 1 st decle contans slaughterng processors wth a relatve mark-down close to zero, ndcatng almost compettve behavour, the last decle comprses the food processors wth a relatvely hgh degree of non-compettve behavour. Table 4.3 provdes estmates of the relatve mark-down for the fruts and vegetables sector. The overall mean of the relatve mark-down for the EU s 0.10, ndcatng small devatons from compettve behavour. The dstrbuton s narrow and skewed toward smaller values. Moreover, the dfferences between the frst and last decles are not so pronounced as compared to slaughterng. That s, the estmate ndcates some degree of market power for fruts and vegetables processors; however, market mperfectons are qute small as compared to the slaughterng sector. COMPETE WORKING PAPER 21

22 Table 4.3 Relatve mark-down Fruts and vegetables Country Statstcal characterstcs of relatve mark-down 1 st 9 th 1 st 3 rd Mean Std.Dev Mn. Max. Decle Decle Quartle Quartle Cases EU Austra Belgum Bulgara Czech Republc Germany Denmark Estona Span Fnland France Unted Kngdom Greece Hungary Italy Ireland NA NA NA NA NA NA NA NA NA Latva Lthuana Netherlands Poland Portugal Romana Sweden Slovena Slovaka Source: own calculaton The market envronment n ndvdual member countres dffers slghtly. The overall mean of the relatve mark-down s n the nterval 0.04 to However, country dfferences are more pronounced f we take nto consderaton the dstrbuton of the relatve mark-down. It holds for all countres that the frst decle comprses companes wth nearly compettve behavour, where the relatve mark-down s close to zero. On the other hand, the last decle s characterzed by consderable non-compettve behavour or degree of market power, as the case may be. Ths holds true prmarly for Belgum, Fnland, France, the Netherlands and Romana. COMPETE WORKING PAPER 22

23 Table 4.4 Relatve mark-down Dary Country Statstcal characterstcs of relatve mark-down Mean Std.Dev Mn. Max. 1 st Decle 9 th Decle 1 st Quartle 3rd Quartle Cases EU Austra Belgum Bulgara Czech Republc Germany Denmark Estona Span Fnland France Unted Kngdom Greece Hungary Italy Ireland Latva Lthuana Netherlands Poland Portugal Romana Sweden Slovena Slovaka Source: own calculaton Table 4.4 presents the relatve mark-down for the EU dary sector. As opposed to the slaughterng and fruts and vegetables sectors, the overall mean of the relatve markdown for the EU common market s qute small. The mean, 0.07, together wth the narrow dstrbuton and skewness toward small values ndcate that the behavour of dary processors s almost compettve wth respect to farmers. In other words, we found only mnor market mperfectons. The last decle suggests that there are only a few companes wth a consderable degree of non-compettve behavour. The dfferences among EU member states are margnal. The mean of the relatve mark-down s, n the majorty of cases, n the nterval 0.05 to A hgher mean can only be found n the Czech Republc (0.081), France (0.075), Portugal (0.078) and Romana (0.105). The frst decle n all countres s very close to zero. The last decle n the majorty of countres s around 0.10, ndcatng that the majorty of dary companes have a small degree of non-compettve behavour. Bulgara (0.19), the Czech Republc (0.17) and Romana (0.26) can be consdered exceptons, wth a hgher relatve mark-down COMPETE WORKING PAPER 23

24 n the last decle. That s, we can fnd severe market mperfectons n these countres as compared to other EU member states. Table 4.5 Relatve mark-down Mllng Country Statstcal characterstcs of relatve mark-down Mean Std.Dev Mn. Max. 1 st Decle 9 th Decle 1 st Quartle 3 rd Quartle Cases EU Austra NA NA NA NA NA NA NA NA NA Belgum Bulgara Czech Republc Germany Denmark Estona Span Fnland France Unted Kngdom Greece Hungary Italy Ireland Latva Lthuana Netherlands Poland Portugal Romana Sweden Slovena Slovaka Source: own calculaton Fnally, Table 4.5 provdes statstcal characterstcs of the estmated relatve mark-down for the mllng sector. As n the case of the dary sector, the overall mean s small, The dstrbuton, whch s narrow and skewed toward smaller values, suggests that the market envronment s not far from compettve behavour on the EU common market. The dfferences among EU member states are also margnal, as n the case of the dary sector. The mean of the relatve mark-down s around 0.06 n the majorty of cases. Only Romana has a slghtly hgher overall mean of the relatve mark-down, The frst decle s very close to zero n all countres. The last decle s around 0.11 n the majorty of countres, ndcatng that the majorty of dary companes have a low degree of noncompettve behavour. Romana s agan an excepton, wth the mllng companes n the last decle havng a substantal degree of market power. That s, we cannot fnd severe market mperfectons n the mllng sector n the majorty of EU member countres. COMPETE WORKING PAPER 24

25 The development of the relatve mark-down nslaughterng sector has a rather stochastc trend. The relatve mark-down for the EU dd not change sgnfcantly between 2003 and The ftted trend functon shows a weak postve trend for 12 countres. However, the changes are margnal n the majorty of cases. The same holds true for the other 12 member states wth an estmated negatve trend. The negatve trend s weak n the majorty of cases, except for Estona. That s, the results suggest that the slaughterng producers dd not sgnfcantly change the degree of non-compettve behavour durng the analysed perod. The estmated trend functons of the relatve mark-down n the fruts and vegetables sector suggest a rather constant trend for the entre EU as well as for Bulgara, the Czech Republc, Denmark, France, Latva, Lthuana and Slovaka. The other countres experenced ether a weak postve or a negatve trend n the majorty of cases. A sgnfcant ncrease n relatve mark-down can only be observed n Slovena. On the other hand, Belgum, the Netherlands and Sweden experenced a sgnfcant decrease n the relatve mark-down. For the dary and mllng sectors we cannot observe any sgnfcant change n the relatve mark-down on the EU level. The same holds true for the majorty of EU member states. However, some exceptons can be found. In the dary sector, the Czech Republc and Slovaka experenced a sgnfcant ncrease n the relatve mark-down (except for 2012, where we have a small number of observatons). On the other hand, a sgnfcant decrease n the relatve mark-down can be observed n France and Lthuana. A sgnfcant ncrease followed by a decrease almost to the ntal values took place n Estona, Portugal and Romana. In the mllng sector, a sgnfcant ncrease n the relatve mark-down can be observed n Bulgara, Estona and Latva, and a decrease n Unted Kngdom, Ireland and Lthuana. 4.2 Mark-up model Table 4.6 provdes a parameter estmate of the mark-up model for the slaughterng, fruts and vegetables, dary and mllng sectors n the 24 EU member states. As n the case of the mark-down model, all the ftted parameters are hghly sgnfcant. Moreover, the hgh sgnfcance of the coeffcents on unobservable fxed management suggests that the chosen specfcaton approxmates well the estmated relatonshp, and that heterogenety among frms s an mportant characterstc of the food processng sector. COMPETE WORKING PAPER 25

26 Table 4.6 Parameter estmates Slaughterng Means for random parameters Coeffcent on unobservable fxed management Varable Coeff. SE P [ z >Z*] Varable Coeff. SE P [ z >Z*] Const Alpha_m Tme Tme Output Output Labour Labour Materal Materal Alpha_mm Sgma Lambda Fruts and vegetables Means for random parameters Coeffcent on unobservable fxed management Varable Coeff. SE P [ z >Z*] Varable Coeff. SE P [ z >Z*] Const Alpha_m Tme Tme Output Output Labour Labour Materal Materal Alpha_mm Sgma Lambda Dary Means for random parameters Coeffcent on unobservable fxed management Varable Coeff. SE P [ z >Z*] Varable Coeff. SE P [ z >Z*] Const Alpha_m Tme Tme Output Output Labour Labour Materal Materal Alpha_mm Sgma Lambda Mllng Means for random parameters Coeffcent on unobservable fxed management Varable Coeff. SE P [ z >Z*] Varable Coeff. SE P [ z >Z*] Const Alpha_m Tme Tme Output Output Labour Labour Materal Materal Alpha_mm Sgma Lambda Source: own calculaton COMPETE WORKING PAPER 26

27 The ftted parameters show that the output and labour nputs have a postve mpact on the revenue share n all sectors. On the other hand, materal nputs determne negatvely the revenue share. Ths result corresponds wth the estmates of the mark-down model. The tme varable has a negatve sgn n slaughterng, fruts and vegetables and mllng and s not statstcally sgnfcant n the dary sector. That s, the revenue share decreases over tme n slaughterng, fruts and vegetables and mllng. The unobservable heterogenety component (management) contrbutes postvely to the revenue share n slaughterng and dary, and negatvely n the fruts and vegetables and mllng sectors. Whereas the postve mpact n slaughterng and dary s acceleratng, the negatve mpact n fruts and vegetables decelerates over tme. If we focus only on the relaton between management and the mark-up component, then an ncrease n the heterogenety component leads to an ncrease n the mark-up component or noncompettve behavour n fruts and vegetables, and to a decrease n the mark-up component n the slaughterng and dary sectors, respectvely. Moreover, hgher output leads to a larger mark-up component n slaughterng and to a smaller mark-up component n the dary and mllng sectors for a gven level of management. The coeffcent on unobservable fxed management for output s not sgnfcant n the fruts and vegetables sector. Labour determnes postvely the mark-up component n slaughterng and negatvely n other sectors for a gven level of management. Fnally, the materal nputs have a negatve mpact on the mark-up component n slaughterng and a postve mpact n other sectors, except for dary, n whch t s not statstcally sgnfcant. As n the case of the mark-down model, these results correspond to our expectatons only partally. The heterogenety (management) component as a measure of the qualty of output and nputs, as well as a measure of good practses, was found to be a source of noncompettve behavour n the fruts and vegetables and mllng sectors. The estmates revealed a postve relaton between the heterogenety component and mark-up component n these sectors. Moreover, the assumpton about the postve relaton between greater sze (or market share n general) and market mperfectons or non-compettve behavour are not met by all coeffcents on unobservable management, as the case may be. However, the postve relaton generally holds. If we regress the relatve mark-up on the output or nput quanttes, we can fnd a sgnfcant postve relatonshp. To sum up, the hypothess about the postve relatonshp between the degree of non-compettve behavour and company sze cannot be rejected; however, t s sgnfcantly determned by the sources of market power. The mpact of tme on market mperfectons s postve n fruts and vegetables and negatve n slaughterng and dary. It s not sgnfcant n the mllng sector. As we concluded for the mark-down model, ths could be a sgn of ncreasng compettveness on the EU food processng common market n the slaughterng and dary sectors. Fnally, the parameter λ s hghly sgnfcant, and greater than one n all sectors. Ths means that the varaton n the mark-up component u t s more pronounced than the varaton n the random component v t. That s, the estmates ndcate that dfferences n noncompettve behavour among food processors are mportant characterstcs of the analysed sectors n the EU. COMPETE WORKING PAPER 27

28 Table 4.7 provdes the estmated relatve mark-up for the EU and ts 24 member states. Snce the relatve mark-up (MU) s n the nterval MU ϵ (0,1), where 0 ndcates compettve behavour and market mperfectons ncrease wth ncreasng MU, the estmated overall mean of the relatve mark-up for the EU slaughterng common market, 0.09, ndcates some degree of non-compettve behavour n the EU slaughterng output market. The dstrbuton of the mark-up s relatvely narrow, wth standard devaton of 0.08, and also a lttle skewed toward smaller values. The frst decle ndcates that 10 % of producers have a rather low mark-up. On the other hand, the last 10 % of producers reach a relatve mark-up greater than 0.14, ndcatng consderable non-compettve behavour. Table 4.7 Relatve mark-up Slaughterng Country Statstcal characterstcs of relatve mark-up Mean Std.Dev Mn. Max. 1 st Decle 9 th Decle 1 st Quartle 3 rd Quartle Cases Total Austra Belgum Bulgara Czech Republc Germany Denmark Estona Span Fnland France Unted Kngdom Greece Hungary Italy Ireland Latva Lthuana Netherlands Poland Portugal Romana Sweden Slovena Slovaka Source: own calculaton The overall means of the relatve mark-up dffer among the ndvdual member states. However, the dfferences are small. The means of the relatve mark-up can be found n the nterval 0.07 to Producers n Austra (0.08), Germany (0.08), Estona (0.08), France (0.08), Unted Kngdom (0.08), Greece (0.08), Italy (0.08), Ireland (0.07), Poland COMPETE WORKING PAPER 28

29 (0.08), Portugal (0.08), Sweden (0.08) and Slovena (0.08) exercse less market power, on average, compared to producers n Bulgara (0.13), Hungary (0.10), the Netherlands (0.11) and Romana (0.15). The dstrbuton of relatve mark-up s narrow n all countres, and a lttle skewed toward small values. However, sgnfcant dfferences between the 1 st and 9 th decles are seen n most EU countres. The largest dfferences can be found n Bulgara, Hungary, the Netherlands and Romana. The hgh value of the relatve mark-up n these countres ndcates the exstence of large market mperfectons. Table 4.8 provdes the estmated relatve mark-up for European processors of fruts and vegetables. In ths sector, the estmated overall mean of the relatve mark-up s Smlarly to slaughterng, ths ndcates the exstence of market mperfectons or some degree of non-compettve behavour. The dstrbuton of the mark-up s relatvely narrow, wth standard devaton of 0.08, and also a lttle skewed toward smaller values. The frst decle ndcates that 10 % of producers have a rather low mark-up (lower than 0.05). On the other hand, the last 10 % of producers reach a relatve mark-up greater than That ndcates a consderable degree of non-compettve behavour. Table 4.8 Relatve mark-up Fruts and vegetables Country Statstcal characterstcs of relatve mark-up Mean Std.Dev Mn. Max. 1 st Decle 9 th Decle 1 st Quartle 3rd Quartle Cases EU Austra Belgum Bulgara Czech Republc Germany Denmark Estona Span Fnland France Unted Kngdom Greece Hungary Italy Ireland NA NA NA NA NA NA NA NA NA Latva Lthuana Netherlands Poland Portugal Romana Sweden Slovena Slovaka Source: own calculaton COMPETE WORKING PAPER 29

30 The overall means of the relatve mark-up dffer among the ndvdual member states. The lowest mean value of the relatve mark-up can be found n Austra (0.09), whle the hghest value was reached n Romana (0.21). Nonetheless, a relatve mark-up wth a value lower than the EU mean can be found n Belgum (0.10), the Czech Republc (0.10), Germany (0.09), Denmark (0.09), France (0.10), Unted Kngdom (0.10), Italy (0.10), Lthuana (0.09), the Netherlands (0.10), Poland (0.10 and Sweden (0.10). The dstrbuton of the relatve mark-up s qute narrow n the analysed states and slghtly skewed toward small values. However, sgnfcant dfferences between the 1 st and 9 th decles exst n most EU countres, especally n Bulgara, Estona, Romana, Slovena and Slovaka. In these countres, market mperfectons are stronger as compared to other EU countres. Table 4.9 presents the relatve mark-up for European dary processors. In ths sector, the estmated overall mean of the relatve mark-up s Ths s the hghest value of the relatve mark-up EU mean as compared to other sectors. Comparng wth the results of the mark-down analyss, we can conclude that producers n the dary sector have Table 4.9 Relatve mark-up Dary Country Statstcal characterstcs of relatve mark-up Mean Std.Dev Mn. Max. 1 st Decle 9 th Decle 1 st Quartle 3rd Quartle Cases EU Austra Belgum Bulgara Czech Republc Germany Denmark Estona Span Fnland France Unted Kngdom Greece Hungary Italy Ireland Latva Lthuana Netherlands Poland Portugal Romana Sweden Slovena Slovaka Source: own calculaton COMPETE WORKING PAPER 30

31 a stronger poston or non-compettve behavour n the output market than n the raw materals nput market,.e. n prce negotatons wth retalers than wth farmers. The dstrbuton of the mark-up s relatvely narrow, wth a standard devaton of However, sgnfcant dfferences between the 1 st and 9 th decles can be found. The frst decle ndcates that 10 % of producers have rather low market power (lower than 0.05). On the other hand, the last 10 % of producers reach a relatve mark-up hgher than That s, the results agan ndcate a consderable degree of non-compettve behavour on the dary output market. The overall means of the relatve mark-up dffer among the ndvdual member states. The means of the relatve mark-up can be found n the nterval 0.08 to Producers n Bulgara (0.09), Unted Kngdom (0.08), Lthuana (0.09) and Sweden (0.08) exercse a lower degree of non-compettve behavour, on average, as compared to producers n Austra (0.15), Fnland (0.15), Hungary (0.16) and Portugal (0.16). The frst decle s very close to zero n Bulgara and Romana. However, t s much hgher n Austra (0.11) and the Netherlands (0.11). It ndcates the exstence of sgnfcant market mperfectons n the dary market n these countres. The spread between the frst and last decles s more pronounced n the dary sector as compared to the slaughterng and fruts and vegetables sectors. Greater dfferences can be found prmarly n Bulgara, Span, France, Hungary, Italy, Portugal and Romana. Fnally, Table 4.10 contans the estmated relatve mark-up for the mllng sector. The overall mean s 0.10, whch also ndcates the exstence of some degree of non-compettve behavour n ths sector on the EU common market. The dstrbuton s agan narrow and a lttle skewed toward the left. The frst decle suggests that 10 % of producers have a rather low mark-up. On the other hand, the last 10 % of producers reach a relatve mark-up hgher than 0.18, ndcatng a consderable degree of non-compettve behavour. The overall means of the relatve mark-up dffer among the ndvdual member states. However, the dfferences are small. The means of the relatve mark-up can be found n the nterval 0.07 to Producers n Belgum (0.09), Germany (0.08), Estona (0.08), Span (0.09), Fnland (0.08), France (0.09) and Slovaka (0.09) exercse smaller market mperfectons, on average, compared to producers n Bulgara (0.14), the Czech Republc (0.13), Denmark (0.12), Greece (0.12) and the Netherlands (0.14). There are sgnfcant dfferences between the 1 st and 9 th decles n most EU countres. The largest dfferences can be found n Bulgara, the Czech Republc, Denmark, Greece, Lthuana, the Netherlands, Poland and Romana. Smlarly to the dary sector, the spread between the 1 st and 9 th decles s more pronounced as compared to slaughterng and fruts and vegetables processng. The majorty of EU member countres experenced a rather stochastc trend of the relatve mark-up n the slaughterng sector. The relatve mark-up for the EU does not change sgnfcantly between 2003 and The ftted trend functon show a weak postve trend for 11 countres. However, the changes are margnal n the majorty of cases, except for Fnland and Latva. The same holds true for the other 13 member states wth an estmated negatve trend. The negatve trend s weak n the majorty of cases. That s, the results suggest that the slaughterng producers dd not change sgnfcantly the degree of non-compettve behavour durng the analysed perod. COMPETE WORKING PAPER 31

32 Table 4.10 Relatve mark-up Mllng Country Statstcal characterstcs of relatve mark-up Mean Std.Dev Mn. Max. 1 st Decle 9 th Decle 1 st Quartle 3 rd Quartle Cases EU Austra NA NA NA NA NA NA NA NA NA Belgum Bulgara Czech Republc Germany Denmark Estona Span Fnland France Unted Kngdom Greece Hungary Italy Ireland Latva Lthuana Netherlands Poland Portugal Romana Sweden Slovena Slovaka Source: own calculaton The estmated trend functons of the relatve mark-up n the fruts and vegetables sector suggest a rather constant trend for the entre EU. A postve trend was estmated for 16 EU member states. However, the changes n the relatve mark-up are agan rather small. Belgum, Estona, Hungary, Portugal and Slovena are exceptons. Whereas Belgum, Portugal and Slovena experenced a sgnfcant ncrease n the relatve mark-up, Estona and Hungary were characterzed by a decrease n the degree of non-compettve behavour n the fruts and vegetables output processng market. We cannot observe any sgnfcant change n the relatve mark-up n the dary sector on the EU level. The same holds true for the majorty of EU member states. However, some exceptons can be found. Hungary and Sweden experenced a sgnfcant ncrease n the relatve mark-up. On the other hand, a sgnfcant decrease n the relatve mark-up could be observed n Portugal. COMPETE WORKING PAPER 32

33 The developments of the relatve mark-up n the mllng sector are characterzed by a stochastc trend n many member states. On the EU level, we cannot observe any sgnfcant changes. The ftted trend functon suggests a weak postve trend. The same holds true for 17 EU member states, where we estmated a weak postve trend n the relatve markup. Hungary and Latva are exceptons, wth a sgnfcant ncrease n the relatve mark-up. 5 Conclusons The estmated mark-down model revealed some degree of non-compettve behavour n the nput food processng market for all analysed sectors,.e. slaughterng, fruts and vegetables, dary and mllng. The degree of market mperfectons dffers among the sectors. Whereas the overall mean of the relatve mark-down for the EU slaughterng common market s , for fruts and vegetables t s , for dary and for mllng That s, the EU slaughterng common market s charactersed by sgnfcantly greater market mperfectons as compared to the dary and mllng sectors, n partcular. Moreover, snce the relatve mark-down s n the nterval zero to one (zero ndcatng no market mperfectons or n general compettve behavour, as the case may be, and a postve value of the relatve mark-down representng non-compettve behavour), the estmated overall means show qute small market mperfectons on the EU nput food processng markets. Ths especally holds true for the dary and mllng sectors. The dstrbuton of the relatve mark-down s relatvely narrow n all sectors, and skewed toward smaller values. Sgnfcant dfferences between the frst and last decle were revealed n slaughterng, ndcatng low market mperfectons for the frst 10 % of producers, but a consderable degree of non-compettve behavour for the last 10 % of slaughterng producers. The dfferences among the producers n fruts and vegetables, and especally n the dary and mllng sectors, are not so pronounced. Furthermore, there were sgnfcant dfferences among EU member countres. In slaughterng, the means of the relatve mark-down could be found n the nterval 0.05 to Bulgara, the Czech Republc, Denmark, Estona, Unted Kngdom, Greece, Ireland, Latva, Lthuana, Poland, Portugal, Romana, Slovena and Slovaka are charactersed by lower market mperfectons as compared to the EU average. On the other hand, Austra, Belgum, Germany, Fnland, France and Italy are countres havng a mean of the relatve mark-down hgher than the EU average. The overall means of the relatve mark-down n fruts and vegetables are n the nterval 0.04 to Belgum, Fnland, France, the Netherlands and Romana are countres wth a relatvely hgh degree of market mperfectons. The dfferences among EU member states are margnal n the dary and mllng sectors. In dary, the mean of the relatve mark-down s n the nterval 0.05 to 0.07 n the majorty of cases. The mean of the relatve mark-down n the mllng sector s around 0.06 n the majorty of cases; only Romana s an excepton, wth a mean of the relatve markdown of COMPETE WORKING PAPER 33

34 Fnally, the development of the relatve mark-down s charactersed by a rather stochastc trend. Ths holds for the majorty of countres n all analysed sectors. However, the relatve mark-down for the EU does not change sgnfcantly between 2003 and The results suggest that the slaughterng producers dd not change sgnfcantly the degree of non-compettve behavour durng the analysed perod. Only margnal changes can be found n other sectors. However, some exceptons can be found. In fruts and vegetables, a sgnfcant ncrease n the relatve mark-down can be observed n Slovena. On the other hand, Belgum, the Netherlands and Sweden experenced a sgnfcant decrease n the relatve mark-down. In the dary sector, the Czech Republc and Slovaka are charactersed by a sgnfcant ncrease n the relatve mark-down, and France and Lthuana by a decrease n the relatve mark-down. In the mllng sector, a sgnfcant ncrease n the relatve mark-down can be observed n Bulgara, Estona and Latva, and a decrease n Unted Kngdom, Ireland and Lthuana. The results of the ftted mark-up model suggest that market mperfectons on the output market are not so pronounced for the slaughterng sector, wth an overall mean of However, the degree of market mperfectons s hgher for the output market n the dary (0.1210) and mllng (0.1016) sectors as compared to the nput market. The fruts and vegetables sector has almost the same mean for the output as for the nput market (0.1061). The dstrbuton of the relatve mark-up s agan relatvely narrow n all sectors and skewed toward smaller values. Moreover, sgnfcant dfferences between the frst and last decle n all sectors were revealed by the estmate. However, the dfferences among producers n the slaughterng sector are not so pronounced. The overall means of the relatve mark-up dffer among the ndvdual member states n slaughterng. However, the dfferences are small. The means of the relatve mark-up can be found n the nterval 0.07 to Producers n Austra, Germany, Estona, France, Unted Kngdom, Greece, Italy, Ireland, Poland, Portugal, Sweden and Slovena exercse less market power, on average, as compared to producers n Bulgara, Hungary, the Netherlands and Romana. The largest dfferences among the slaughterng producers can be found n Bulgara, Hungary, the Netherlands and Romana. The hgh value of the relatve mark-up n these countres ndcates the exstence of strong market mperfectons. In fruts and vegetables, the lowest mean value of the relatve mark-up can be found n Austra (0.09), and the hghest value was reached n Romana (0.21). Nonetheless, a value of the relatve mark-up lower than the EU mean can be found n Belgum, the Czech Republc, Germany, Denmark, France, Unted Kngdom, Italy, Lthuana, the Netherlands, Poland and Sweden. On the other hand, strong market mperfectons exst n Bulgara, Estona, Romana, Slovena and Slovaka as compared to other EU countres. The means of the relatve mark-up are n the nterval 0.08 to 0.18 for dary producers. Bulgara, Unted Kngdom, Lthuana and Sweden exercse a lower degree of non-compettve behavour, on average, as compared to producers n Austra, Fnland, Hungary and Portugal. The frst decle s very close to zero n Bulgara and Romana. However, t s much hgher n Austra and the Netherlands. Ths ndcates the exstence of sgnfcant market COMPETE WORKING PAPER 34

35 mperfectons n the dary market n these countres. The spread between the frst and last decles s more pronounced n the dary sector as compared to the slaughterng and fruts and vegetables sectors. Greater dfferences can be found prmarly n Bulgara, Span, France, Hungary, Italy, Portugal and Romana. The dfferences n the overall means of the relatve mark-up n the mllng sector are not so pronounced among the countres. Means of the relatve mark-up can be found n the nterval 0.07 to Producers n Belgum, Germany, Estona, Span, Fnland, France and Slovaka exercse, on average, smaller market mperfectons compared to producers n Bulgara, the Czech Republc, Denmark, Greece and the Netherlands. Sgnfcant dfferences between the 1 st and 9 th decles exst n most EU countres. The largest dfferences can be found n Bulgara, the Czech Republc, Denmark, Greece, Lthuana, the Netherlands, Poland and Romana. Smlarly to the dary sector, the spread between the 1 st and 9 th decles s more pronounced as compared to slaughterng and fruts and vegetables processng. The development of the relatve mark-up s agan rather stochastc, and changes are only margnal n the majorty of cases. That s, the results suggest that the producers dd not sgnfcantly change the degree of non-compettve behavour durng the analysed perod on the output market between the years However, some exceptons can be found. In fruts and vegetables, Belgum, Portugal and Slovena experenced a sgnfcant ncrease n the relatve mark-up, and Estona and Hungary are charactersed by a decrease n the degree of non-compettve behavour. Hungary and Sweden experenced a sgnfcant ncrease and Portugal a sgnfcant decrease n the relatve mark-up on the dary output market. In mllng, a sgnfcant ncrease n the relatve mark-up can be found n Hungary and Latva. COMPETE WORKING PAPER 35

36 6 References Acharya, R. N., Knnucan, H. W., Caudll, S. B. (2011) Asymmetrc farm-retal prce transmsson and market power: a new test, Appled Economcs, 43 (30), pp Álvarez, A., Aras, C., Greene, W. (2003) Fxed Management and tme nvarant techncal effcency n a random coeffcent model. Workng Paper, Department of Economcs, Stern School of Busness, New York Unversty, p. 10. Álvarez, A., Aras, C., Greene, W. (2004) Accountng for unobservables n producton models: management and neffcency. Economc Workng Papers at Centro de Estudos Andaluces E2004/72, Centro de Estudos Andaluces, p. 18 Bakucs, L. Z., Fertö, I., Hockmann, H., Perekhozhuk, O. (2009) Market power on the edge? An analyss of the German and Hungaran hog markets. Agrarwrtschaft, 58 (8), pp Bresnahan, T. F. (1982) The Olgopoly Soluton Concept s Identfed, Economcs Letters, 10, pp Bresnahan, T. F. (1989) Emprcal Stude sof Industres wth Market Power. In: Schmalensee, R., Wllg, R. D. Handbook of Industral Economcs 2, North-Holland, Amsterdam, The Netherlands. Pp Bugues, P. A. and Rey, P. (2004) The Economcs of Anttrust and Regulaton n Telecomuncatons: Perspectves for the New European Regulatory Framework, Edward Elgar Publshng Lmted, UK. Compte, O., F. Jenny and Rey, P. (2002) Capacty Constrants, Mergers and Colluson, European Economc Revew, 46 (1), pp Fudenberg, D., and Trole, J. (1990) Moral Hazard and Renegotaton n Agency Contracts, Econometrca, Econometrc Socety, 58, pp Hockmann, H., and Vönekm É. (2009) Colluson n the Hungaran market for raw mlk. Outlook on Agrculture, 38 (1), pp Ivald, M., Jullen B., Rey P., Seabrght P., Trole, J. (2003) The Economcs of Tact Colluson, Fnal Report for DG Competton, EC, IDEI, Toulouse. Jondrow, J. et al. (1982) On the Estmaton of Techncal Ineffcency n the Stochastc Fronter Producton Functon Model, Journal of Econometrcs, 19, pp Kumbhakar, S. C., Baardsen, S., Len, G. (2012) A New Method for Estmatng Market Power wth an Applcaton to Norwegan Sawmllng. Revew of Industral Organzaton, 40, pp McCorrston, S. (2002) Why should mperfekt competton matter to agrcultural economsts?, European Revew of Agrcultural Economcs, 29, pp Morrson Paul, C. J. (2000) Cost Economes and Market Power: The Case of The U.S. Meat Packng Industry, Workng Paper No COMPETE WORKING PAPER 36

37 Muth, M. K., and Wohlgenant, M. K. (1999) Measurng the Degree of Olgopsony Power n the Beef Packng Industry n the Absence of Marketng Input Quantty Data, Journal of Agrcultural and Resource Economcs, 24, pp Perekhozhuk, O., and Grngs, M. (2006) Measurng the degrese of market power n the Ukranan mlk processng ndustry. In Curtss, J., Balmann, A., Dautzenberg, K., Happe, K.. Studes on the Agrcultural and Food Sector n Central and Eastern Europe, IAMO, Halle, Germany. pp Perekhozhuk, O., Hockmann, H., Bakucs, L. Z., Fertö, I. (2011) Identfcaton of market power n the Hungaran dary ndustry: A plant-level analyss. Paper presented at the EAAE Congress Change and Uncertanty, Challenges for Agrculture, Food and Natural Resources, ETH Zurch, Zurch, Swtzerland, August 30 to September 2, Scalco, P. R., and Braga, M. J. (2014) Measurng the Degree of Olgopsony Power n the Brazlan Raw Mlk Market, Internatonal Food and Agrbusness Management Revew, 17 (2), pp Schroeter, J., and Azzam, A. (1990) Measurng market power n mult-product olgopoles: the US meat ndustry, Appled Economcs, 22, pp Trole, J. (1988) The Theory of Industral Organzaton. MIT Press. Cambrdge and London Welwta, A., and Azzam, A. (1996) Identfyng Implct Colluson under Declnng Output Demand, Journal of Agrcultural and Resource Economcs, 21 (2), pp COMPETE WORKING PAPER 37

38 Project nformaton Ttle: Fundng: Duraton: Objectve: Coordnator: Consortum: Contact: Webste Internatonal comparsons of product supply chans n the agr-food sectors: determnants of ther compettveness and performance on EU and nternatonal markets (COMPETE) Collaboratve research project (small or medum-scale focused research project), FP-7-KBBE , total EU contrbuton s 2,422,725 01/10/ /09/2015 (36 months) The objectve of the COMPETE project s to gan a more comprehensve vew on the dfferent elements whch contrbute to the compettveness of the European agr-food supply chan n order to provde better targeted and evdence based polces on the EU as well as on the domestc level. The project nvestgates selected determnants of compettveness lke polcy nterventons and the busness envronment, productvty n agrculture and food processng, the functonng of domestc and nternatonal markets, the choce of governance structures, and nnovatve actvtes n food processng. The research results wll enable a congruent, coherent and consstent set of polcy recommendatons amng at mprovng compettveness of European product supply chan. IAMO, Germany, Prof. Henrch Hockmann 16 Partners from 10 European countres. COMPETE brngs together academcs, trade bodes, NGOs, agrcultural co-operatve, ndustry representatve advsory servces. In addton, the project s supported by the group of socetal actors, ncorporatng farmer, food processng and consumer assocatons, provdng n-depth knowledge on the agrfood sector and speedng up the achevement of the project goals. compete@amo.de COMPETE Project Coordnaton: Lebnz Insttute of Agrcultural Development n Transton Economes (IAMO) Theodor-Leser-Str.2 D Halle Telephone: Fax: Emal: compete@amo.de