PRODUCTIVE EFFICIENCY OF PORTUGUESE VINEYARD REGIONS

MARTA-COSTA A., MARTINHO V., SANTOS M., Regional Science Inquiry, Vol. IX, (2), 2017, pp. 97-107 97 PRODUCTIVE EFFICIENCY OF PORTUGUESE VINEYARD REGIONS Ana MARTA-COSTA Corresponding auhor. Assisan Professor. Universiy of Trás-os-Mones e Alo Douro (UTAD) and Cenre for Transdisciplinary Developmen Sudies (CETRAD), Porugal, www.uad.p amara@uad.p Víor MARTINHO Coordinaor Professor. Agriculura School, Polyechnic Insiue of Viseu, Porugal vdmarinho@esav.ipv.p. Micael SANTOS Research fellow. Universiy of Trás-os-Mones e Alo Douro (UTAD) and Cenre for Transdisciplinary Developmen Sudies (CETRAD) micaels@uad.p Absrac The overall globalizaion in wine indusry and he search for susainabiliy of he secor has increased compeiion which highlighs he imporance of produciviy gains. The purpose of his paper is o analyse he producive efficiency of he viiculure secor for he Poruguese regions, over he period 1989 o 2007, wih daa from he EUFADN, using boh a deerminisic and sochasic approach. The resuls show an increase of Technical Efficiency (TE) when used he sochasic froniers analysis (SFA) in all regions, while he daa envelopmen analysis (DEA) approach hrough he Malmquis inde reveals a sabilizaion of TE. Keywords: Efficiency, produciviy, Poruguese regions, viiculure. JEL classificaion: C6, Q1 1. Inroducion Over he las decades he wine indusry has been subjec o an inensive globalizaion and inernaional compeiion, a fac ha poses boh challenges and opporuniies o he wine regions, which are compelled o adop innovaive sraegies. Despie recen impressive performance of he New World wine counries, boh in oupu and epors, he European Union (EU-28) coninues o be, in 2014, world s leader in wine producion, occupying almos 50% of vineyards area worldwide and responsible for around 65% of he wine producion by volume and rade (GAIN, 2015). Neverheless, he recen rend has been for he EU-28 vine growing area (of jus under 3.5 million ha in 2013) o decline, due o shrinking margins and EU subsidies paid o farmers o uproo heir vines. The EU s Common Agriculure Policy (CAP) aims o increase he compeiiveness of he wine indusry, mainaining he bes radiional pracices, reinforcing he rural social fabric and preserving environmenal susainabiliy. The winemaking secor plays also a key role in he Poruguese economy. Porugal is he 11h wine producer and he 9h eporer in value and volume in he world (OIV, 2016). Half of he oal naional wine producion is epored, which represens nearly 2% of naional epors (IVV, 2013). Porugal s 2014 wine producion was 3.62% of he EU-28 (589 million lires) and he 2015 grape growing area was 217,000 ha (GAIN, 2015 and OIV, 2016). The counry is divided ino 14 main high qualiy wine producing region. In recen decades he grapevine/wine secor has been modernized, creaing sric regulaions o guaranee wine characerisics, making his secor generally more compeiive on he global marke. However, he increase of compeiiveness imposed by globalizaion has led new challenges and hreas for he survival of he firms in he wine secor (Sellers-Rubio e al., 2016). The grape growers should achieve beer levels of efficiency and improve he performance of heir firms who makes he sudies on he producive efficiency more imporan. The producers always aspire a increasing he efficiency of heir producion process, bu hey do no always succeed in opimizing i. The research of reasons behind firm

98 MARTA-COSTA A., MARTINHO V., SANTOS M., Regional Science Inquiry, Vol. IX, (2), 2017, pp. 97-107 inefficiency is very imporan o correc fragiliies and o improve efficiency in he producion cone. The producion efficiency esimaion regarding differen vineyard sysems plays an imporan role in he field of economic cone, as i allows managemens o reach high levels of performance resricing resources, minimizing undesired inpus or opimizing desired oupus. However, organizaions compeiiveness allied o efficiency analysis mus be allocaed also as a key ool in improving social wellbeing and promoing susainable developmen (Mara-Cosa, 2010a, 2010b; Mara-Cosa and Silva, 2013) as he concep evolves he parsimony use of economic resources in order o reach cos minimizaion, oupu and profi maimizaion (Sampaio, 2013). Given he coninuous pressure on vineyard producs margins, improvemens in organisaions economic efficiency can boos he financial siuaion of boh farms and vineyard-dependen communiies. The aim of his paper is o analyse he Technical Efficiency (TE) of he vineyard secor in Poruguese regions, using boh a deerminisic and sochasic approach, over he period 1989 o 2007, using daa from he European Farm Accounancy Daa Nework (EUFADN, 2017). 2. Producive Efficiency of Viiculure Secor: A Lieraure Review The concep of producive efficiency has been decomposed ino echnical and allocaive efficiency, by Farrel (1957). The firs occurs when given a cerain level of inpus, he Decision Making Uni (DMU) is able o produce he maimum level of oupus or, given a cerain level of oupu, he DMU is able o minimize he level of inpu (Fleming e al., 2014). The allocaive efficiency reflecs he abiliy of a firm o use he inpus in opimal proporions, given heir respecive prices o minimize he cos or maimize he revenue (Aparicio e al., 2013). Increasing producion efficiency on he wine secor calls for a beer undersanding of he key facors driving he performance of vineyard sysems. This could be a difficul ask due o he heerogeneous characerisics of he used producion sysems (Sellers-Rubio e al., 2016). For insance, in he case of Porugal, he farms are composed by several blocks and each block represen a differen variey of grapes/vines and managemen. The muliple effecs of he grapevine physiology and he grapevine-environmen-agronomy ineracions are widely recognised, and have implicaions on yield and berry qualiy, which in urn deermine he major drivers of he vineyard efficiency. Relaed o he heerogeneous characerisics of he used producion sysems, for eample Moreira e al. (2011), in heir sudy of Chilean wine grape producers, classified four caegories of planing vine sysems: simple cordon, double cordon, pergola and oher raining sysem, which affec he producive efficiency of farms. The pergola sysem was he mos producive, while he simple and double cordon were less producive. Regarding o TE he oher raining sysems ehibi a highes level, followed by simple cordon, pergola and double cordon. Oher sudies relae efficiency levels boh o microeconomic facors, such as farm and farmer characerisics (Conradie e al., 2006; Henriques e al., 2009); grape variey (Coelli and Sanders, 2013; Manevska-Tasevska, 2013); diversificaion and specialisaion (Henriques e al., 2009; Coelli and Sanders, 2013); irrigaion (Conradie e al., 2006; Henriques e al., 2009), and o macroeconomic facors, such as developmens in he financial sysem; he qualiy of human capial, and per capia wine consumpion (József and Péer, 2014). The value and volume of grapes or wine are he main variables idenified by he lieraure regarding he oupus of he vineyard sysems. According o Sellers-Rubio e al. (2016) and Sellers-Rubio and Más-Ruiz (2015) he oupu in moneary erms reveals he abiliy of wine producer o ransform inpus ino wine oupus value or value of grapes produced and i represens he economic concep of efficiency. Alhough he quaniy of oupu in volume corresponds o he echnical concep of efficiency and i is evaluaed by he wine producer abiliy o ransform some inpus ino lires of wine or quaniy of grapes produced. In oher words, some farms are concerned o produce as much as possible a he lowes price, while oher prefers o produce less and wih higher qualiy o sell wines wih a higher price. According o Coelli and Sanders (2013) variable of oupu esimaed in quaniy erms does no ake ino accoun differences in qualiy. The producion efficiency esimaion of wine secor, in erms of quaniy, is presen in he sudies of Tóh & Gál (2014), Moreira e al.

MARTA-COSTA A., MARTINHO V., SANTOS M., Regional Science Inquiry, Vol. IX, (2), 2017, pp. 97-107 99 (2011) and Freias (2014). Alhough, he producion efficiency esimaion, in erms of value, giving aenion no only o price of each wine or grape produced bu also o he quaniy, was developed by Sellers-Rubio e al. (2016), Sellers-Rubio & Más-Ruiz (2015), Aparicio e al. (2013), Vidal e al. (2013), Brandano e al (2012), and Henriques e al. (2009). These wo ways of efficiency concepualizaions of he farms is of grea imporance in supporing famers decisions o adap heir vineyards and producion sysems and open discussion regarding he model o pursue for he fuure, someimes associaed wih he availabiliy of he daa. The sudies wih values of oupu use variables such us sales revenue, profi volume and earnings (Sellers-Rubio e al. 2016; Sellers-Rubio & Máz-Ruiz, 2015; and Brandano e al., 2012), oal oupu in value (Henriques e al., 2009) and value of domesic and foreign sales (Aparicio e al., 2013; Vidal e al., 2013). While when oupu are in volume he used variables are he quaniy of wine producion (Tóh & Gál, 2014) and quaniy of grapes produced (Moreira e al., 2011; Freias, 2014). For inpus, we found hree main caegories of variables for he sudy of he vine-growing sysem efficiency: labour, capial and land. However, inermediae consumpion can be an imporan inpu like Freias (2014) sudied. Moreover, Coelli and Sanders (2013) idenify irrigaion for inpus oo, bu his variable is imporan only for dryland farms. Some sudies idenify only wo inpus (labour and capial), such as Sellers-Rubio e al. (2016), Sellers- Rubio & Más-Ruiz (2015), Freias (2014) and Brandano e al. (2012). Bu, Tóh & Gál (2014), Moreira e al. (2011) and Henriques e al. (2009) added he used land while Aparicio e al. (2013) used also he number of vinegrowers. Oher sudies (Aparicio e al., 2013; Vidal e al., 2013) oped by he lonely inpu of land. Also, he same variable could be measured by differen ways. The labour inpu is sudied by he number of employees (Sellers-Rubio e al., 2016; Sellers-Rubio & Más-Ruiz, 2015) or by agriculural employmen (Tóh & Gál, 2014), labour cos (Moreira e al., 2011; Brandano e al., 2012) or by work hours (Freias, 2014; Henriques e al., 2009). The land inpu variable seems o be more consensual and are measured by he used area of land in hecares. For he capial inpu, he papers use some variables such as equiy level of winery (Sellers-Rubio e al., 2016), funds of company (Sellers-Rubio & Más-Ruiz, 2015), level of deb (Sellers-Rubio e al., 2016; Sellers-Rubio e al., 2015), ne agriculural capial sock (Tóh & Gál, 2014) and capial like machinery coss and oher fied asses (Moreira e al., 2011; Freias, 2014; Brandano e al. 2012; and Henriques e al., 2009). In addiion o hese inpus, some sudies also include he inermediae consumpion wih he vegeal producion such he coss wih pesicides and ferilizers (Moreira e al. 2011; Freias, 2014; and Henriques e al., 2009). These mulipliciy of he variables used for producive efficiency esimaion inensifies he discussion abou he differen resuls ha could be gahered and how could hey influence he decision of he farmers and oher sakeholders of he indusry. Besides he main groups of variables of oupus and inpus, he lieraure refers also o oher variables regarding he inefficiency of he farms. Some eamples of hese variables are he access o finance, planing vine sysem, ype of wine, grape qualiy, ype of firm, employmen rae, average of wages paid, gross domesic produc and ohers (Moreira e al., 2011; Fuensanana e al., 2015; Tóh & Gal, 2014). In general, he reflecion of he lieraure of producion efficiency sudies shows ha here are some variables ha are consensual and common in several works, as land area, labour and capial (inpus). However, here are oher variables ha can affec he efficiency in differen ways bu here are no consensual. The grea diversiy of variables ha have already been used, makes he sudy of produciviy and efficiency very comple and complicaed o esimae. Despie his, hese sudies are very imporan for wine-growing producers because he efficiency is a goal ha hey have in mind. Empirical sudies have adoped wo main alernaive approaches, ofen based eiher on (non-parameric, non-sochasic) mahemaical programming models or on (sochasic, parameric) economeric models. Boh are very useful o suppor decision-making because hey give an indicaion of wha he main characerisics of farm are and how some of he problems idenified may be solved (Silva e al., 2013). According o Coelli e al. (2005) eis four mehods: leas-squares economeric producion models; oal facor produciviy indices; daa envelopmen analysis (DEA); and sochasic

100 MARTA-COSTA A., MARTINHO V., SANTOS M., Regional Science Inquiry, Vol. IX, (2), 2017, pp. 97-107 froniers analysis (SFA). The firs wo models assume ha all firms are echnically efficien, alhough he las wo are he mehods mos used o esimae efficiency especially a one poin in ime. While he mehod leas-squares economeric producion models and sochasic froniers involve he economeric esimaion of parameric funcions, he mehod oal facor produciviy indices and DEA do no (Coelli e al, 2005). The DEA and SFA are he mehods mos used in lieraure of producive efficiency. DEA esimae he efficiency fronier and he disance of producion fronier of he DMUs o he efficien fronier give us a measure of inefficien (Sellers-Rubio e al., 2016; Sellers-Rubio & Más Ruiz, 2015; Brandano e al., 2012). DEA is a non-parameric mehodology based on linear programing echniques and is no necessary o include specific funcional forms for he producion funcion (relaionship beween inpus and oupus) o esablish he efficien fronier (Sellerrs-Rubio & Más Ruiz, 2015; Brandano e al., 2012; Cullinane e al., 2006; Henriques e al., 2009; Moreira e al., 2011). A DMU is efficien in DEA analysis if no oher DMU is able o produce more oupus from he same inpus or is able o produce he same oupus wih less inpus (Sellers-Rubio & Más Ruiz, 2015). Moreover, a DEA can sudy a framework wih muliple inpu-oupu focuses on a virual single inpu-oupu srucure (Brandano e al., 2012; Cullinane e al., 2006). Furhermore, DEA considers specificaion error as minimal, so no consider random shocks or measuremen errors and all facor are due o inefficiency (Cullinane e al., 2006). When panel daa is available, he DEA allows he calculaion of he Malmquis inde, which esimaes he s in producive efficiency during he analysed period (Fare e al., 1994). SFA is an alernaive approach of DEA and was inroduced simulaneously by Aigner e al. (1977) and Meeusen and van Den Broeck (1977). This approach as some advanages, such as include measuremen errors and random shocks, analyze he srucure and invesigae he deerminans of producer performance. However, o have his srucure and o invesigae deerminans i is needed a more solid economic heory (Cullinane e al., 2006). Moreover, he SFA needs o have a funcional form, so here is a greaer risk of imposing a priori assumpions abou producion echnology and i is difficul o verify he precise specificaion of he error srucure (Cullinane e al., 2006). The Sochasic Producion Froniers (SPF) is used by several aricles like Moreira e al. (2011); Tóh & Gal (2014) and Mourão and Marinho (2016). Alhough, he DEA is used by sudies such as Sellers-Rubio e al. (2016), Sellers-Rubio e al. (2015), Freias (2014), Vidal e al. (2013), Fuensana e al. (2015), Aparicio e al. (2013) and Brandano e al. (2012). 3. Daa and Empirical Model The daabase from EU (EUFADN, 2017) is available for he period 1989 o 2007 and for he five regions of Porugal (Enre Douro e Minho and Beira Lioral; Trás-os-Mones and Beira Inerior; Ribaejo e Oese; Alenejo and Algarve; and Açores and Madeira), which allows a panel daa, despie he aggregaed available informaion in an average value. Descripive saisics for he daa used in his analysis are presened in Table 1. Descripive saisics Table 1. Descripive saisics of variables for regions from 1989 o 2007 Wine and Vineyards Inpus for all he aciviies of he farms grapes area (Ha) Labour Toal specific (Euros) (Hours) coss (Euros) Average farm capial (Euros) Obs. 95 95 95 95 95 Mean 1939 0.864 3370 5536 35703 Sd. Dev. 1291 0.634 455 3610 18989 Min 276 0.040 2141 1729 19216 Ma 5539 2.120 4754 17839 106245 Source of daa: EUFADN (2017) Of sressing ha he sample farms is around 500 and 1000 in Enre Douro e Minho and Beira lioral, Trás-os-Mones and Beira inerior, and Ribaejo e Oese (in his region d o 200-500 afer he year 2000). In Alenejo and Algarve, and Açores and Madeira he sample is around 200 and 500.

MARTA-COSTA A., MARTINHO V., SANTOS M., Regional Science Inquiry, Vol. IX, (2), 2017, pp. 97-107 101 In his work we sudied he SPF funcion proposed by Aigner e al. (1977) and Meeusen and van Den Broeck (1977). Alhough, we used an eension of he original model esablished by Baese and Coelli (1992), ha is ypically used when panel daa is available o eplain he in TE (Baese and Coelli, 1995). The used model can be represened as: (1) where (1) is a Cobb and Douglas (1928) funcion, being he Yi he wine and grape p roducion in euros of he i-h region in he -h year; α is he consan; β is he unknown parameers o be esimaed, being X1 he vineyard area in hecares; X2 represens he labour in hours; X3 he average farm capial in euros; and X4 he oal specific coss in euros; vi is he random error and ui is he non-negaive random variable, associaed wih echnical inefficiency in producion of firms in he indusry involved. The seleced inpu variables of his model are he mos consensual in he lieraure ha affecs he TE of he vineyard secor. When his model is esimaed, he TE is given by: (2) The calculaions of he model (1) and (2) were performed wih resource o he SFA (imevarying decay) and using Saa (2017) sofware. Time-varying decay models allow for s in he TE over he considered period and he perinence of hese models is verified hrough he ea values (Baese and Coelli, 1992). Ea refers o ime-varying inefficiency effecs and posiive values signify ha he firms improve heir TE over ime. In his sudy we also used he Malmquis (1953) inde, which was inroduced by Caves e al. (1982) wih resource o DEA model (Fare e al., 1994) using he DEA Program (DEAP) 2.1 version. The DEAP allows he applicaion of Malmquis DEA mehods o panel daa o calculae indees of oal facor produciviy (TFP) divided in efficiency and echnical. The efficiency represens he deviaions of bes pracice fronier, while he echnical reflecs he fronier shif over ime, so he efficiency reflecs he caching-up effec, while echnical reflecs fronier shif effec (Sellers-Rubio e al., 2016). The Malmquis inde is esimaed using disance funcions and hrough he available panel daa i allows he sudy of he s in efficiency on differen regions of Porugal. An inpu-oriened model was used o obain an efficien uni based on a proporional decrease of is inpu, while he oupus proporions remain und (Coelli e al., 2005). The Malmquis inde quanifies he in oal facor produciviy (TFP) beween wo daa poins by calculaing he disance raio of each poin relaive o a common fronier. According o Grifell-Tajé and Lovell (1996), for a given uni, he (produc-oriened) inde of Malmquis TFP beween periods (base period) and +1 is given by: + 1 d (y 1) d (y 1) + + + + M,+ 1(y+ + y, ) = + 1 d (y, ) d (y, ) (3) M, + 1(y+ + y, ) is he geomeric mean of wo Malmquis indees (raios beween disance funcions), adoping he firs as reference echnology of period and he second one of period + 1. A value of he Malmquis inde greaer, equal or less han one indicaes he occurrence of growh, sagnaion or decline of TFP. where (1) is a Cobb and Douglas (1928) funcion, being he Yi he wine and grape p roducion in euros of he i-h region in he -h year; α is he consan; β is he unknown parameers o be esimaed, being X1 he vineyard area in hecares; X2 represens he labour in hours; X3 he average farm capial in euros; and X4 he oal specific coss in euros; vi is he random error and ui is he non-negaive random variable, associaed wih echnical inefficiency in producion of firms in he indusry involved. The seleced inpu variables of his model are he mos consensual in he lieraure ha affecs he TE of he vineyard secor. When his model is esimaed, he TE is given by: (2) The calculaions of he model (1) and (2) were performed wih resource o he SFA (imevarying decay) and using Saa (2017) sofware. Time-varying decay models allow for s in he TE over he considered period and he perinence of hese models is verified 1/2

102 MARTA-COSTA A., MARTINHO V., SANTOS M., Regional Science Inquiry, Vol. IX, (2), 2017, pp. 97-107 hrough he ea values (Baese and Coelli, 1992). Ea refers o ime-varying inefficiency effecs and posiive values signify ha he firms improve heir TE over ime. In his sudy we also used he Malmquis (1953) inde, which was inroduced by Caves e al. (1982) wih resource o DEA model (Fare e al., 1994) using he DEA Program (DEAP) 2.1 version. The DEAP allows he applicaion of Malmquis DEA mehods o panel daa o calculae indees of oal facor produciviy (TFP) divided in efficiency and echnical. The efficiency represens he deviaions of bes pracice fronier, while he echnical reflecs he fronier shif over ime, so he efficiency reflecs he caching-up effec, while echnical reflecs fronier shif effec (Sellers-Rubio e al., 2016). The Malmquis inde is esimaed using disance funcions and hrough he available panel daa i allows he sudy of he s in efficiency on differen regions of Porugal. An inpu-oriened model was used o obain an efficien uni based on a proporional decrease of is inpu, while he oupus proporions remain und (Coelli e al., 2005). The Malmquis inde quanifies he in oal facor produciviy (TFP) beween wo daa poins by calculaing he disance raio of each poin relaive o a common fronier. According o Grifell-Tajé and Lovell (1996), for a given uni, he (produc-oriened) inde of Malmquis TFP beween periods (base period) and +1 is given by: + 1 d (y 1) d (y 1) + + + + M,+ 1(y+ + y, ) = + 1 d (y, ) d (y, ) (3) M, + 1(y + + y, ) is he geomeric mean of wo Malmquis indees (raios beween disance funcions), adoping he firs as reference echnology of period and he second one of period + 1. A value of he Malmquis inde greaer, equal or less han one indicaes he occurrence of growh, sagnaion or decline of TFP. 4. Resuls The able 2 presens he resuls of he model (1). Is analysis shows ha he wine and grape producions are mainly eplained by he vineyard area and by he average farm capial (saisically significan). These resuls reveals he imporance of he effors made o modernize he farms, namely wih he srucural financial suppors from he Communiy Suppor Frameworks applied in he Poruguese agriculural secor. The oher inpus (labour and specific coss) are no saisically significan. Probably his siuaion occurs because hese inpus reflec all he aciviies of he farm, insead only from he vineyard aciviy. All parameers have a posiive signal, which means ha he increase of he inpus (average farm capial, labour and vineyard area) allows an increase of he producion, wih he ecepion of he specific coss ha have an inverse relaionship wih he producion of wine and grapes. However, he values of he parameers show he perinence of he model, namely he resuls for he parameer gamma. They reveal a model consan o he scale (1.082) which means ha he increase in he quaniy of use of he inpus deermines a proporional increase of he quaniy of he oupu. 1/2

MARTA-COSTA A., MARTINHO V., SANTOS M., Regional Science Inquiry, Vol. IX, (2), 2017, pp. 97-107 103 Table 2. Resuls considering SFA of producion funcion (ime-varying decay), across he period 1989-2007 and for five Poruguese regions, wih he logarihm of wine and grape producions as dependen variable Variables Parameers and saisical Consan 3.484 (1.070) Logarihm of vineyard area 0.550* (9.330) Logarihm of labour 0.033 (0.090) Logarihm of oal specific coss -0.052 (-0.400) Logarihm of average farm capial 0.447* (2.310) /mu 0.100 (1.220) /ea 0.109* (4.170) /lnsigma2-2.244* (-14.570) /ilggamma -2.935* (-2.160) sigma2 0.105 Gamma 0.050 sigma_u2 0.005 sigma_v2 0.100 Noe: *, saisically significan a 5%. Source of daa: EUFADN (2017). The able 3 presens he resuls for TE (ime-varing decay) for he Poruguese regions over he period 1989-2007. Table 3. Resuls for TE (ime-varying decay), across he period 1989-2007 and for five Poruguese regions Year Enre Douro e Minho and Beira Lioral Trás-os- Mones and Beira inerior Ribaejo and Oese Alenejo and Algarve Açores and Madeira Annual mean 1989 0.851 0.427 0.460 0.221 0.491 0.490 1990 0.865 0.466 0.498 0.258 0.528 0.523 1991 0.877 0.504 0.535 0.297 0.564 0.555 1992 0.889 0.541 0.571 0.336 0.598 0.587 1993 0.900 0.576 0.605 0.376 0.631 0.617 1994 0.909 0.610 0.637 0.416 0.661 0.647 1995 0.918 0.642 0.667 0.456 0.690 0.675 1996 0.926 0.672 0.696 0.494 0.717 0.701 1997 0.934 0.700 0.722 0.532 0.742 0.726 1998 0.940 0.726 0.747 0.567 0.765 0.749 1999 0.946 0.750 0.770 0.602 0.786 0.771 2000 0.952 0.773 0.791 0.634 0.806 0.791 2001 0.956 0.794 0.810 0.665 0.824 0.810 2002 0.961 0.813 0.828 0.693 0.841 0.827 2003 0.965 0.831 0.844 0.720 0.856 0.843 2004 0.968 0.847 0.859 0.745 0.870 0.858 2005 0.972 0.861 0.873 0.768 0.883 0.871 2006 0.974 0.875 0.885 0.789 0.894 0.884 2007 0.977 0.887 0.896 0.809 0.904 0.895 Annual mean 0.931 0.700 0.721 0.546 0.740 0.727 TE 2007 /TE 1989 148 2,077 948 3,661 841 827 The resuls for he esimaed TE (able 3) demonsrae he increase of inde over he period 1989-2007, for all Poruguese regions. Enre Douro e Minho and Beira Lioral have he higher TE (annual mean of 0.931) while Alenejo and Algarve have he lower values (annual mean of 0.546). However, he mos efficien group of regions - Enre Douro e Minho and Beira Lioral - only has a TE of 1.148 imes for 2007 greaer han 1989, while he mos inefficien

104 MARTA-COSTA A., MARTINHO V., SANTOS M., Regional Science Inquiry, Vol. IX, (2), 2017, pp. 97-107 group of regions - Alenejo and Algarve - has a TE 3.661 imes greaer in 2007, relaed o 1989. The resuls regarding he DEA approach hrough he Malmquis inde are in able 4. Table 4. Malmquis inde resuls of annual and region means Malmquis Inde summary of annual means Year Efficiency Technical Pure TE Scale efficiency Toal facor produciviy 1990 0.999 0.657 0.999 1.000 0.656 1991 0.909 0.854 0.909 1.000 0.776 1992 0.863 1.462 1.068 0.808 1.261 1993 1.257 1.123 1.016 1.238 1.412 1994 1.016 0.681 1.016 1.000 0.692 1995 0.911 0.865 0.911 1.000 0.787 1996 0.960 2.021 1.097 0.875 1.940 1997 1.095 0.969 0.958 1.143 1.061 1998 1.042 0.699 1.042 1.000 0.728 1999 1.001 0.754 1.001 1.000 0.754 2000 0.862 2.467 0.967 0.891 2.126 2001 1.138 0.838 1.014 1.122 0.954 2002 0.984 0.823 0.984 1.000 0.809 2003 1.034 0.532 1.034 1.000 0.550 2004 0.894 3.280 0.935 0.956 2.934 2005 1.112 0.687 1.063 1.046 0.764 2006 1.007 0.885 1.007 1.000 0.891 2007 0.974 0.987 0.974 1.000 0.960 Mean 0.998 1.000 0.998 1.000 0.998 Accumulaed 1990/2007 0.973 0.998 0.972 1.000 0.965 Malmquis Inde summary of firm means Region Efficiency Technical Pure TE Scale efficiency Toal facor produciviy Enre Douro e Minho and Beira Lioral 1.000 0.833 1.000 1.000 0.833 Trás-os-Mones and Beira inerior 0.998 0.909 0.998 1.000 0.907 Ribaejo e Oese 1.000 1.008 1.000 1.000 1.008 Alenejo and Algarve 0.997 1.056 0.997 1.000 1.052 Açores and Madeira 0.998 1.240 0.998 1.000 1.237 Mean 0.998 1.000 0.998 1.000 0.998 On he one hand, he analysis reveals ha, in an annual perspecive, he TFP increased in 1992 (because of he echnical ), 1993 (in consequence of increases of all indees), 1996 (srong increase in he echnical ), 1997 (increase in he efficiency, derived from he scale efficiency ) and srongly in 2000 and 2004 (because of greaer increases in he echnical ). On he oher hand, TFP decreased in he oher years, due o he echnical, and in 2007 due o he efficiency. Considering he mean and accumulaed values over he analysed period and for he five Poruguese regions, i is observed ha he TFP decreased slighly because of small decreases in he efficiency, derived from he efficiency, more precisely of he pure echnical efficiency. In any case, he TFP shows an increasing rend during he period in erms of regional analysis. Ribaejo and Oese, Alenejo and Algarve, and Açores and Madeira improved heir TFP derived from he increases in he echnical. However, he TFP of Enre Douro e

MARTA-COSTA A., MARTINHO V., SANTOS M., Regional Science Inquiry, Vol. IX, (2), 2017, pp. 97-107 105 Minho and Beira Lioral, and Trás-os-Mones and Beira Inerior regions decrease due essenially o he decrease of echnical. The join analyse of he previous ables (3 and 4) emphasize differen resuls. While he SFA (able 3) shows he increasing of efficiency, he Malmquis inde (able 4) shows ha he efficiency produciviy inde do no over he period 1989-2007. The gains of efficiency, hrough SFA (able 3) decrease in las years, from 0.810 (2001) o 0.895 (2007), in annual mean values. Alhough, he Malmquis inde (able 4) had more loses in he efficiency, wih indees of 0.863 (1992), 0.862 (2000) and 0.894 (2004), precisely he years closes o he mos imporan CAP reforms. In erms of regional analysis, while he SFA shows ha he Enre Douro e Minho and Beira Lioral are he mos echnical efficien region (able 3) and wih some growh, he Malmquis inde shows ha TFP decreases over he period 1989 o 2007, due o he echnical (able 4). However, he Alenejo and Algarve is he mos inefficien regions hrough SFA (able 3), bu wih he greaes progress. The Malmquis inde emphasizes his siuaion, once i reveals a posiive in TFP for he same regions. 5. Conclusions The esimaion of produciviy efficiency of Poruguese regions over he period 1989-2007 conduc o he differen resuls when is used SFA and DEA hrough Malmquis inde. Based on he firs approach all regions improve TE over he period 1989-2007, while i is noed a decrease when calculaed he Malmquis inde or TFP due o he efficiency. I means ha some farms have difficulies in approaching he bes fronier of producion. In regional erms, SFA shows he improved echnical progress for all Poruguese regions, especially in he mos inefficien regions, such as Alenejo and Algarve, while Malmquis inde reveals ha only wo groups (Açores and Madeira; and Alenejo and Algarve) enhanced he TFP and oher wo regions decreased he inde (Enre Douro e Minho and Beira lioral; and Trás-os-Mones and Beira Inerior). All hese s in Malmquis inde are derived from echnical ha reveals ha some regions had echnological progress and ohers had loses in efficiency due o he non-modernizaion of is producion echnologies. 6. References Aigner, D., Lovell, C. K., and Schmid, P. 1977. Formulaion and esimaion of sochasic fronier producion funcion models, Journal of Economerics, 6 (1): 21-37. Aparicio, J., Borras, F., Pasor, J. T., and Vidal, F. 2013. Accouning for slacks o measure and decompose revenue efficiency in he Spanish Designaion of Origin wines wih DEA. European Journal of Operaional Research, 231 (2): 443-451. Baese, G. E., and Coelli, T. J. 1992. Fronier producion funcions, echnical efficiency and panel daa: wih applicaion o paddy farmers in India. Journal of Produciviy Analysis, 3: 153-169. Baese, G. E., and Coelli, T. J. 1995. A model for echnical inefficiency effecs in a sochasic fronier producion funcion for panel daa. Empirical Economics, 20: 325-332. Brandano, M. G., Deoo, C., and Vannini, M. 2012. Comparaive Efficiency of Producer Cooperaives and Convenional Firms in a Sample of Quasi-Twin Companies (No. 201228). Sardinia: Cenre for Norh Souh Economic Research, Universiy of Cagliari and Sassari. Caves, D. W., Chrisensen, L. R., and Diewer, W. E. 1982. The economic heory of inde numbers and he measuremen of inpu, oupu, and produciviy. Economerica: Journal of he Economeric Sociey, 50 (6): 1393-1414. Cobb, C. W., and Douglas, P. H. 1928. A Theory of Producion. American Economic Review, 18 (Supplemen): 139-165. Coelli, T. J., Rao, D. S. P., O'Donnell, C. J., and Baese, G. E. 2005. An Inroducion o Efficiency and Produciviy Analysis. Springer Science & Business Media. Coelli, T., and Sanders, O. 2013. The echnical efficiency of wine grape growers in he Murray- Darling Basin in Ausralia. In Giraud-Héraud, E. and Pichery, M. C. (eds), Wine Economics. Quaniaive Sudies and Empirical Applicaions. Hampshire (UK): Palgrave Macmillan, 231-249. Conradie, B., Cookson, G., and Thirle, C. 2006. Efficiency and farm size in Wesern Cape grape producion: pooling small daases. Souh African Journal of Economics, 74: 334-343. Cullinane, K., Wang, T. F., Song, D. W., and Ji, P. 2006. The echnical efficiency of conainer pors: comparing daa envelopmen analysis and sochasic fronier analysis. Transporaion Research Par A: Policy and Pracice, 40 (4): 354-374.

106 MARTA-COSTA A., MARTINHO V., SANTOS M., Regional Science Inquiry, Vol. IX, (2), 2017, pp. 97-107 EUFADN: European Union Farm Accounancy Daa Nework, 2017. European Union Farm Accounancy Daa Nework. hp://ec.europa.eu/agriculure/rica/ (accessed January 24, 2017). Fare, R., Grosskopf, S., and Lovell, C. K. 1994. Producion Froniers. Cambridge Universiy Press. Farrell, M. J. 1957. The measuremen of producive efficiency. Journal of he Royal Saisical Sociey, Series A (General), 120 (3): 253-290. Fleming, E., Mouner, S., Gran, B., Griffih, G., and Villano, R. 2014. The New World challenge: Performance rends in wine producion in major wine-eporing counries in he 2000s and heir implicaions for he Ausralian wine indusry. Wine Economics and Policy, 3 (2): 115-126. Freias, R. 2014. Sobre a eficiência dos países produores de uvas para vinho na União Europeia: uma aproimação DEA em duas eapas. Masers Disseraion, Caholic Universiy of Porugal. Fuensana, M. J. R., Sancho, F. H., and Marco, V. S. 2015. In vino verias: compeiive facors in wine-producing indusrial disrics. Invesigaciones Regionales, (32), 149. GAIN: Global Agriculural Informaion Nework. 2015. Global Agriculural Informaion Nework, Wine Annual Repor and Saisics 2015. USDA Foreign Agriculural Services, GAIN Repor Number: IT1512. Grifell-Taje, E., and Lovell, C. K. 1996. Deregulaion and produciviy decline: The case of Spanish savings banks. European Economic Review, 40 (6): 1281-1303. Henriques, P.D.S., Carvalho, M.L.S., and Fragoso, R.M.S. 2009. Technical efficiency of Poruguese wine farms. New Medi, 8 (1): 4-9. Herrera, B., Gerser-Benaya, M., and Knierim, A. 2016. Sakeholders percepions of susainabiliy measuremen a farm level. Sudies in Agriculural Economics, 118: 131-137. József, T., and Péer, G. 2014. Is he New World more efficien? Facors influencing echnical efficiency of wine producion. Sudies in Agriculural Economics, 116: 95-99. Malmquis, S. 1953. Inde numbers and indifference surfaces. Trabajos de esadísica, 4 (2): 209-242. Manevska-Tasevska, G. 2013. Produc assormen and he efficiency of farms. In Giraud-Héraud, E. and Pichery, M. C. (eds), Wine Economics. Quaniaive Sudies and Empirical Applicaions. Hampshire, UK: Palgrave Macmillan, 250-265. Mara-Cosa, A., and Silva, E. (eds.) 2013. Mehods and Procedures for Building Susainable Farming Sysems: Applicaion in he European Cone. Neherlands: Springer Ediions. Mara-Cosa, A. A. 2010a. Agriculura Susenável III: Indicadores. Revisa de Ciências Agrárias, 33 (2): 90-105. Mara-Cosa, A. A. 2010b. Susainabiliy Sudy for he Rearing of Bovine Livesock in Mounainous Zones. New Medi, 9 (1): 4-12. Meeusen, W., and van Den Broeck, J. 1977. Efficiency esimaion from Cobb-Douglas producion funcions wih composed error. Inernaional economic review, 18 (2): 435-444. Moreira, V. H., Troncoso, J. L., and Bravo-Urea, B. E. 2011. Technical efficiency for a sample of Chilean wine grape producers: A sochasic producion fronier analysis. Ciencia e Invesigación Agraria, 38 (3): 321-329. Mourão, P. R., and Marinho, V. D. 2016. Scoring he efficiency of Poruguese wine epors an analysis recurring o Sochasic Fronier Models. Ciência Téc. Viiv., 31 (1): 1-13. OIV: Inernaional Organisaion of Vine and Wine. 2016. OIV Saisical Repor on World Viiviniculure: World Viiviniculure Siuaion. hp://www.oiv.in/public/medias/5029/worldviiviniculure-siuaion-2016.pdf (accessed January 2016). PTFADN: Poruguese Farm Accounancy Daa Nework. 2017. Poruguese Farm Accounancy Daa Nework. hp://www.gpp.p/inde.php/rica/rede-de-informacao-de-conabilidades-agricolas-rica (accessed January 24, 2017). Sampaio, A. 2013. Review of fronier models and efficiency analysis: a parameric approach. In Mendes, A., Silva, E. and Sanos, J. (eds), Efficiency Measures in he Agriculural Secor, Springer Ediion, 13-35. Sellers-Rubio, R., and Más-Ruiz, F. J. 2015. Economic efficiency of members of proeced designaions of origin: sharing repuaion indicaors in he eperience goods of wine and cheese. Review of Managerial Science, 9 (1): 175-196. Sellers-Rubio, R., Alampi Soini, V., and Menghini, S. 2016. Produciviy growh in he winery secor: evidence from Ialy and Spain. Inernaional Journal of Wine Business Research, 28 (1): 59-75. Silva, E., Mendes, A., and Sanos, J. 2013. Efficiency measures in he agriculural secor: he beginning. In Mendes, A., Silva, E. and Sanos, J. (eds), Efficiency Measures in he Agriculural Secor, Springer Ediion, 3-12. Saa. 2017. Daa Analysis and Saisical Sofware. hp://www.saa.com/ (accessed January 24, 2017). Tóh, J., and Gál, P. 2014. Is he New Wine World more efficien?. Sudies in Agriculural Economics, 116 (2): 95-99.

MARTA-COSTA A., MARTINHO V., SANTOS M., Regional Science Inquiry, Vol. IX, (2), 2017, pp. 97-107 107 Vidal, F., Pasor, J. T., Borras, F., and Pasor, D. 2013. Efficiency analysis of he designaions of origin in he Spanish wine secor. Spanish Journal of Agriculural Research, 11 (2): 294-304. Acknowledgemens: This work was suppored by he projec NORTE-01-0145-FEDER-000038 (INNOVINE & WINE Vineyard and Wine Innovaion Plaform) and by European and Srucural and Invesmen Funds in he FEDER componen, hrough he Operaional Compeiiveness and Inernaionalizaion Programme (COMPETE 2020) [Projec No 006971 (UIC/SOC/04011)]; and naional funds, hrough he FCT Poruguese Foundaion for Science and Technology under he UID/SOC/04011/2013.