Double concentration explaining the outstanding increase in Spanish crop production

Double concentration explaining the outstanding increase in Spanish crop production Miguel Martín-Retortillo (Martín-Retortillo, M)1, Ana Serrano (Serrano, A)2 and Ignacio Cazcarro (Cazcarro, I)3 1 Universidad de Alcalá. Faculty of Economics, Business and Tourism, Dept. of Economics, Plaza de la Victoria 2, 28802 Alcalá de Henares, Spain. 2 Universidad de Zaragoza. Agrifood Institute of Aragon (IA2). Faculty of Economics and Business Studies, Dept. of Economic Analysis. Gran Vía 2, 50005 Zaragoza, Spain. 3 ARAID (Aragonese Agency for Research and Development). Agrifood Institute of Aragon (IA2). Faculty of Economics and Business Studies, Dept. of Economic Analysis. Gran Vía 2, 50005 Zaragoza, Spain. Abstract Aim of study: To evaluate the changes in Spanish agricultural production since 1950s in a context of intense transformations in terms of the regional and crop composition. Area of study: Spanish provinces during the second half of the twentieth century. Material and methods: We use index decomposition analysis to evaluate the changes in the value and volume of crop production, as well as the role of product composition and the regional distribution of production. Main results: Spanish agriculture have focused on certain regions in the south or in the east of Spain. Some products like vegetables or fruits have a positive prices and composition effects, encouraging the production in these provinces. Research highlights: We found a ‘double concentration’: Spanish agriculture has increasingly tended to produce high value-added products, such as vegetables, fruit and olive oil. On the other hand, crop production is concentrated in the southern and eastern provinces of Spain. Additional key words: Spain; agricultural sector; agricultural development; index decomposition; regional analysis Abbreviations used: CAP (Common Agricultural Policy); CE (Composition Effect); EEC (European Economic Community); IDA (Index Decomposition Analysis); LMDI (Logarithmic Mean Divisia Index); NRA (Nominal Rate of Assistant); PE (Price Effect); QE (Quantity Effect); RE (Share Effect); SE (Scale Effect) Authors’ contributions: These authors contributed equally to this work. All authors read and approved the final manuscript. Citation: Martín-Retortillo, M; Serrano, A; Cazcarro, I (2020). Double concentration explaining the outstanding increase in Spanish crop production. Spanish Journal of Agricultural Research, Volume 18, Issue 3, e0107. https://doi.org/10.5424/sjar/2020183-15760 Supplementary material: (Table S1 and Figure S1) accompanies the paper on SJAR’s website Received: 19 Sep 2019. Accepted: 02 Jul 2020. Copyright © 2020 INIA. This is an open access article distributed under the terms of the Creative Commons Attribution 4.0 International (CC-by 4.0) License. Competing interests: The authors have declared that no competing interests exist. Correspondence should be addressed to Miguel Martín-Retortillo: [email protected] Spanish Journal of Agricultural Research 18 (3), e0107, 12 pages (2020) eISSN: 2171-9292 https://doi.org/10.5424/sjar/2020183-15760 Instituto Nacional de Investigación y Tecnología Agraria y Alimentaria (INIA) OPEN ACCESS RESEARCH ARTICLE Funding agencies/institutions Project / Grant Ramón Areces Foundation CISP15A3198 Introduction Agricultural production and productivity in Europe and their drivers have been among the main themes of agricultural economics (Hayami & Ruttan, 1985; Grigg, 1992; O’Brien & Prados, 1992; Federico, 2005; Lains & Pinilla, 2009; Alston & Pardey, 2014; Gollin et al., 2014). They developed strongly during the second globalization (the second half of the twentieth century), owing to major transformations in the sector. These included technical innovations that increased dependence on the non-agricultural sectors of the economy, growing market integration of agricultural products and a reduction in the share of the labour and land factors of production (while capital increased its share). The case of Spain is striking for, among other reasons, its steady increase in agricultural production and productivity, especially during the second half of the twentieth century (Clar et al., 2016). Such continuous growth was not seen in other European countries, and since the 1980s agricultural production has stagnated (Martín-Retortillo & Pinilla, 2015). This growth was not spread 2Miguel Martín-Retortillo, Ana Serrano and Ignacio Cazcarro Spanish Journal of Agricultural Research September 2020 • Volume 18 • Issue 3 • e0107 evenly across the Spanish regions, and it also varied by product group. On a massive scale, Spain introduced technical innovations from the non-agricultural sectors of the economy, such as agricultural machinery, chemical fertilizers and hybridization and selection of seeds (Grigg, 1992; Abad & Naredo, 1997; Hassine & Kandil, 2009). Moreover, the development of irrigation boosted agricultural production thanks to the great advances in water infrastructure. Spain’s agricultural mechanization was faster than that of other European countries, despite administrative obstacles to the widespread introduction of tractors in the 1950s (Clar, 2009).1 Thus, Spain converged with the European level of mechanization (Martín & Pinilla, 2015). Second, chemical products such as fertilizers and pesticides were another important non-agricultural input. While the introduction of these chemicals was significant throughout Europe during the twentieth century (and especially after the Second World War), in Spain it was done on a huge scale, so that once again the country converged on European levels. Third, as mentioned above, biological innovations did much to bring about an improvement in agricultural production and productivity. These advances, which began in the final decades of the nineteenth century (Olmstead & Rhode, 2008), intensified in the 1940s and 1950s, focusing on the hybridization and genetic selection of seeds. The most developed countries all benefited from these innovations, but because of their particular climate conditions, the yield increase was lower in the Mediterranean countries than in Western Europe (Pujol, 2011; Harwood, 2018). Finally, the expansion of irrigation was also relevant in a European Mediterranean country like Spain. On the one hand, the technical innovations related to irrigation infrastructure allowed water to be supplied to arid regions: while the area of land irrigated more than doubled on the European continent generally, in Spain it almost tripled in the second half of the twentieth century. This expansion was especially intensive during the 1950s, 1960s and 1970s, when the Franco dictatorship promoted the construction of dams as a pillar of its agrarian policy. Irrigated production as a share of total agricultural production increased from 42.3% to 65.5% between 1955 and 2006 (Pinilla, 2008; Cazcarro et al., 2015a; Martín-Retortillo & Pinilla, 2015). On the other hand, the use of plastic greenhouses, sprinkler 1 Despite the higher increase in agricultural mechanization in Spain, the level of mechanization was lower than the Western European countries in the decades of the 1940 and 1950 (Martín-Retortillo & Pinilla, 2015). 2 The Nominal Rate of Assistance (NRA) tries to approximate to the percentage by which government policies have raised gross returns to farmers above what they would have been without the government’s intervention (or reduced them, if NRA < 0). In the empirical work, the NRA is expressed as a percentage of the undistorted price (Anderson & Valenzuela, 2008). 3 The second period of analysis includes the entrance of Spain in the ECC, with the CAP in place. This involved policies of support in prices (subsidies) to some products such as cereals until around 1991. From 1992 there was a reorientation towards direct aids to maintain farmer’s agrarian income (coupled with production until 2003 and then decoupled). This implies some heterogeneities in the processes of prices creation, in the same fashion that different drivers (institutional, changes in openness’s and demands from foreign markets, livestock demand, etc.) have played a role in prices formation. The deflation used in this article takes into account the general inflation of all the crop products in the sample, and hence it is more a reference which allows to distinguish a differential evolution of relative prices among crops, than a method to capture and disentangle these drivers. and drip irrigation and sandy soils were among the most important technical innovations that allowed an improvement in water productivity (López and Losada, 1999; Cazcarro et al., 2019). Apart from these technical developments, the new institutional framework was also key to the expansion of Spanish agricultural production. Spanish agriculture took a step toward reintegration with the Stabilization and Liberalization Plan of 1959, but it was the country’s 1986 entry to the European Economic Community (EEC henceforth) that provided the main stimulus. In the period from 1960 to 1986, Spanish agri-food exports grew at an annual rate of 5.8%, while imports rose by 4.7%. This export growth coincides with a slight increase in the openness ratio during the 1970s, and with a strong increase from 1985 to the first years of the twenty-first century (Clar et al., 2015). Spain’s entry into the EEC not only eased commercial exchanges, but also modified the level of agricultural protectionism. Intervention in the agricultural sector was relatively low until Spain became a member of the EEC. In fact, the Nominal Rate of Assistance (NRA) was very low during the 1950s, 1960s and early 1970s, compared with other Western European countries.2 Basically, the common agricultural policy (CAP henceforth) guaranteed high prices (generally higher than international prices), which encouraged agricultural protection from the early days of the EEC right up until the MacSharry reform of 1992. Spain’s low level of agricultural protection increased from 1986, when the country joined the EEC. Thus, the NRA grew from approximately 10%, as a percentage of the undistorted price in the first half of the 1980s, to a maximum of 70% by the end of that decade (Anderson & Valenzuela, 2008).3 In sum, the lack of agricultural support and the dictatorial context until the later 1970s both generated an agricultural system based on the productivist model. This model focused on increases in agricultural production, the expansion of irrigation, the large-scale adoption of non-agricultural technical innovations, and modification of the regional and crop patterns of agricultural production. All this, combined with a high number of hours of sunshine and better availability of water resources increased the sector’s potential in the following decades. Since the 1980s, with accession to the EEC and implementation of the CAP, this increase in production has been linked to international markets – especially in the case of those export products Spanish Journal of Agricultural Research September 2020 • Volume 18 • Issue 3 • e0107 3 Double concentration explaining the outstanding increase in Spanish crop production in greatest demand (Clar et al., 2018), such as vegetables, olive oil and wine. With all of that, the objective of this paper is to analyse the changes in Spanish crop production during the second half of the twentieth century and investigate how, where and why it experienced such a sustained and significant growth. Material and methods Data In this section, we go into the main data sources to analyse Spanish agricultural production since the 1950s. First, we collected the physical agricultural production (Qijt expressed in tonnes) from Spain’s Agrarian Statistics Yearbooks for 1955, 1980, 2005 and 2010 (MAPAMA, 1955, 1980, 2005, 2010). We worked with 48 Spanish provinces (see Table S1 [suppl.] for the names of the provinces and the relation to wider administrative regions) and 132 agricultural products. Then we obtained information on the current prices (Pcit expressed in pesetas/tonne) received by farmers for each crop (MAPAMA, 1955, 1980, 2005, 2010). Note that in order to separate the inflation impact and to account for the change in real production and in the value of crops, the data have been deflated, using 1980 as a base year. We calculated price indices for the periods 19551980 and 1980-2005/2010 using the 132 crops included in the sample and taking 1980 as a base year. These indices inform on the average change experienced by the prices of the crops during both time-spans. Thus, they can be considered as a reference that allows evaluating the changes in crop prices over time. The calculated agricultural price indices were around 7.4 and 2.8 for 1955 to 1980 and 1980 to 2005/2010, respectively.4 The prices indices are used to obtain 1955 and 2005/20105 deflated prices (Pit expressed in constant 1980 pesetas), introduced below in Eq. (1) to estimate monetary agricultural production. Hence, the paper focuses on three key moments for the Spanish agricultural sector. First, we look at the mid-twentieth century (1955), just before the 1959 Stabilization and Liberalization Plan mentioned above, which brought an end to Spanish autarky and heralded openness to international markets. This period also preceded the rapid development of hydraulic infrastructures during the middle years of Franco’s dictatorship. Second, as a mid-point we focus on 1980, 4 These indices are very similar to the ones calculated from the agriculture’s value added deflator provided by Prados de la Escosura (2017) that reach 7.6 and 2.5 for 1955 to 1980 and 1980 to 2005/2010, respectively. For the whole economy, the price increase was larger as the indices obtained from the GDP deflator in Prados (2017) are 11.4 and 4.6 for 1955 to 1980 and 1980 to 2005/2010, respectively. 5 We take these years to examine changes up to the present, since there was relatively high variation in precipitations in these years, with 2005 a relatively dry year and 2010 a relatively humid year (1955 and 1980 were both average years). 6 Checking the representativeness of our years, agricultural production of 1955 represents 1.02 on 5 years average focused on that year; the case of 1980 represents 1.04 on 5 years average focused on that year; 2005 represents 0.94 on 5 years average focused on that year; and 2010 represents 1.02 on 5 years average focused on that year. 7 For observations with zero values, we apply the analytical limit strategy when doing the decomposition analysis (Ang & Liu, 2007). when a large number of water infrastructures had already been developed. This moment in time also witnessed the beginning of democracy in Spain and preparations for the country to join the EEC, which triggered a great commercial expansion and important changes for agriculture. Finally, we looked at the beginning of the twenty-first century by averaging a dry (2005) and a humid (2010) year, following Cazcarro et al. (2015b).6 This final period paints a picture of a democratic, globally integrated and more environmentally aware country. Methodology After describing the data sources and characteristics, we explain the methods used for the empirical analysis. To start, we estimated monetary agricultural production (Prodijt expressed in pesetas) for each crop i, province j and year t. We used information on the physical production (Qijt) and prices (Pit) introduced in the previous section, which yields: (1) From Eq. (1) we can also obtain, for each period, the total production by region ( ), the total production by crop ( ) and the total national production ( ). Then, we proceeded to quantify the main forces behind the historical changes in this variable. This allowed us to identify the most significant structural patterns driving trends in the value of agricultural production. With this objective, we used decomposition analysis, which links the change to an aggregate variable (i.e. monetary production in our case) with a set of determinants. Note that the effects derived from decomposition analysis display the changes that would have happened if all the other factors had remained constant. A previous study on agricultural patterns using decomposition analysis for different explanatory factors and different regional and sectoral detail can be found in Cazcarro et al. (2015b). Of the existing decomposition techniques, we utilized IDA (Ang & Zhang, 2000); and more specifically, we concentrated on the Logarithmic Mean Divisia Index (LMDI henceforth). We followed Ang (2015), who suggests using the LMDI-I additive decomposition presented in Ang et al. (2010) when working with a quantity indicator.7 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑄𝑄𝑖𝑖𝑖𝑖𝑖𝑖 𝑃𝑃𝑖𝑖𝑖𝑖 4Miguel Martín-Retortillo, Ana Serrano and Ignacio Cazcarro Spanish Journal of Agricultural Research September 2020 • Volume 18 • Issue 3 • e0107 Note that the whole set of effects presented in this section was calculated for the two sub-periods of analysis (1955 to 1980 and 1980 to 2005/2010). In order to obtain these impacts for the entire period (1955 to 2005/2010), we followed Ang et al. (2010), who stated that the chaining procedure is preferable when more than two years’ worth of data are available, since it takes better account of the ups and downs of the period.8 Thus, we got two sets of decompositions (1950 to 1980 and 1980 to 2005/2010) that are chained, yielding a result for the whole period (1955 to 2005/2010) (Ang, 1994; Ang & Liu, 2007; Ang et al., 2010). Before going into the formulation of the decomposition effects, we need to define the logarithmic mean of monetary production. Using 1 for the final period and 0 for the initial period, we get: (2) After that, following Ang et al. (1998) we proceeded with the decomposition of monetary agricultural production. We first developed the ‘value decomposition’ and then the ‘reallocation decomposition’. In both, their multi-dimensional (crop and province dimensions) and multi-level (different levels of aggregation) character allows an in-depth examination of the long-term agricultural changes and patterns, highlighting their main structural heterogeneities.9 Thus, Eq. (3) represents the change in monetary agricultural production between periods 0 and 1. (3) This change can also be expressed as the sum of the quantity and price effects obtained using the LMDI-I additive decomposition introduced by Ang et al. (1998). This yields Eqs. (4) and (5) for the quantity (QE) and price (PE) effects, respectively: (4) (5) This first decomposition, the ‘value decomposition’, tries to capture the extent to which long-term changes in monetary agricultural production are driven by changes in the volume of production (QE) and by changes in the value (price) of the different crops (PE). Note that although monetary agricultural production has been deflated for the whole crop production, the price effect exists as a result of the price differences among different products and might 8 They literally acknowledge that ‘chaining analysis is preferred because it gives results which are more representative of the true situation, it makes full use of the data available, and it is more flexible in terms of application’. 9 For a review of the importance of the multi-level and multi-dimensional character of IDA, see Ma (2014). be especially informative when working with regional and crop detail. This perspective extends to the role of agricultural expansion and to the specific crops and regions responsible for the increase in the value of Spanish production that has resulted from the growing production of high-value products. Similarly, in our second decomposition, the change in monetary agricultural production between years 1 and 0 can be expressed as the sum of the change in the product shares within provinces (composition effect, CE), the change in provincial composition of production (share effect, RE) and the change in total monetary production (scale effect, SE). =∑∑𝑐𝑐𝑖𝑖𝑖𝑖1∗𝑟𝑟𝑖𝑖1 ∗𝑠𝑠1 𝑖𝑖𝑖𝑖 −∑∑𝑐𝑐𝑖𝑖𝑖𝑖0∗𝑟𝑟𝑖𝑖0 ∗𝑠𝑠0 𝑖𝑖=𝐶𝐶𝐶𝐶+𝑅𝑅𝐶𝐶+𝑆𝑆𝐶𝐶 𝑖𝑖 ∆𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃=𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃1−𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃0=∑∑(𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖1 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖1 ∗𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖1 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃1∗𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃1)− 𝑖𝑖𝑖𝑖 ∑∑(𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖0 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖0 ∗𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖0 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃0∗𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃0) 𝑖𝑖𝑖𝑖 ∆𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃=𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃1−𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃0=∑∑(𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖1 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖1 ∗𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖1 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃1∗𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃1)− 𝑖𝑖𝑖𝑖 ∑∑(𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖0 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖0 ∗𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖0 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃0∗𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃0) 𝑖𝑖𝑖𝑖 =∑∑𝑐𝑐𝑖𝑖𝑖𝑖1∗𝑟𝑟𝑖𝑖1 ∗𝑠𝑠1 𝑖𝑖𝑖𝑖 −∑∑𝑐𝑐𝑖𝑖𝑖𝑖0∗𝑟𝑟𝑖𝑖0 ∗𝑠𝑠0 𝑖𝑖=𝐶𝐶𝐶𝐶+𝑅𝑅𝐶𝐶+𝑆𝑆𝐶𝐶 𝑖𝑖 (6) Again, following Ang et al. (1998), we estimated the CE, RE and SE in the following equations: 𝐶𝐶𝐶𝐶=∑∑𝐿𝐿(𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝑖𝑖𝑖𝑖1 ,𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝑖𝑖𝑖𝑖0 ) 𝑖𝑖ln(𝑐𝑐𝑖𝑖𝑖𝑖1 𝑐𝑐𝑖𝑖𝑖𝑖0)⁡ 𝑖𝑖 (7) 𝑅𝑅𝑅𝑅=∑∑𝐿𝐿(𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖1,𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖0) 𝑖𝑖ln(𝑃𝑃𝑖𝑖1 𝑃𝑃𝑖𝑖0)⁡ 𝑖𝑖 (8) 𝑆𝑆𝑆𝑆 =∑∑𝐿𝐿(𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖1,𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖0) 𝑖𝑖ln(𝑠𝑠1 𝑠𝑠0)⁡ 𝑖𝑖 (9) Accordingly, we can explain the changes in monetary production as a result of variations in: 1) the crop composition of production (CE) that allows us to evaluate the productive change toward higheror lower-value products within regions; 2) the regional redistribution of production (RE) that provides information on the geographical displacement of monetary production between regions; and 3) the scale of monetary production (SE). Hence, the ‘reallocation decomposition’ sheds light on the contribution of structural changes, i.e. on the changes in crop and regional specialization patterns. Results The Spanish agricultural production: a view based on regions and products Table 1 shows the level of production in monetary units and its annual growth rates. We can observe continuous growth in Spanish agricultural production of almost 2% 𝐿𝐿(𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖1,𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖0)=𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖1− 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖0 ln(𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖1)−ln⁡(𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖0) ∆𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 = 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃1−𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃0= ∑∑𝑄𝑄𝑖𝑖𝑖𝑖1𝑃𝑃𝑖𝑖1 𝑖𝑖𝑖𝑖 −∑∑𝑄𝑄𝑖𝑖𝑖𝑖0𝑃𝑃𝑖𝑖0 𝑖𝑖=𝑄𝑄𝑄𝑄+𝑃𝑃𝑄𝑄 𝑖𝑖 𝑄𝑄𝑄𝑄=∑∑𝐿𝐿(𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖1,𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖0) 𝑖𝑖ln(𝑄𝑄𝑖𝑖𝑖𝑖1 𝑄𝑄𝑖𝑖𝑖𝑖0)⁡ 𝑖𝑖 𝑃𝑃𝑃𝑃 =∑∑𝐿𝐿(𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖1,𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖0) 𝑖𝑖ln(𝑃𝑃𝑖𝑖1 𝑃𝑃𝑖𝑖0)⁡ 𝑖𝑖 Spanish Journal of Agricultural Research September 2020 • Volume 18 • Issue 3 • e0107 5 Double concentration explaining the outstanding increase in Spanish crop production annually over more than 50 years. The highest growth was from 1955 to 1980, when the figure increased by 3.3% annually. This growth coincides with the widespread adoption of chemical products and agricultural machinery and the construction of major hydraulic infrastructures during the Franco dictatorship.10 However, growth was slower from 1980 to 2005/2010 (0.7% annually). Even then, however, Spanish agriculture increased its production, in contrast to other European countries, where production remained stable from the mid-1980s. Table 1 and Fig. 1 show that crop production was concentrated in the south and east of Spain, especially in Andalusia and Murcia; meanwhile, the northwest lost importance in terms of agricultural production. If at the beginning of the period the region of Andalusia had a share of 20.2% of Spanish crop production, and 23.5% in 1980, by the end of the period this had risen to 31.9%. Other regions –like Aragon, Extremadura and Murcia– substantially increased their own importance in Spanish crop production. This concentration coincides with a period of vast construction of hydraulic infrastructures during the Franco dictatorship, particularly in the Mediterranean area and the Ebro basin. By contrast, the two regions of 10 There was also some construction of hydraulic infrastructures before the Franco dictatorship, especially from 1911 to 1936. However, the most significant building was in the 1950s, 1960s and 1970s (Herranz, 2004). Castile and Leon and Galicia, both situated in the northwest, lost almost half of their weighting in total crop production. Looking at Fig. 1, the largest increases in Spanish crop production occurred in the provinces of Almeria, Badajoz, Murcia, Seville, Zaragoza, Huesca, Valencia, Cordoba, Granada and Lerida, being many of them in the east, and several in the south of Spain. The case of Almeria –the most south-easterly province and the most arid – is striking. The role of increased irrigation in this water-scarce region (albeit a region with good soil and plenty of sunshine hours) is a dominant explanatory factor in the agricultural revolution that occurred. Here we find a yearly growth in crop production of 6.3% throughout the period. The different specialization of each province produced these differential growths. In order to obtain a complete view, it is necessary also to observe the evolution of production in terms of groups of products; this helps us to understand the changing crop patterns of Spanish agriculture. As Table 2 shows, the biggest growth is to be found in vegetables, which moved from making up 10% of the total production value in 1955 to 32% in 2005/2010. Table 2 also shows increases in the share of Table 1. Monetary production value (prices deflated using the base year 1980) by region (in million pesetas) and annual growth rates Sources: Own elaboration from MAPAMA (1955, 1980, 2005, 2010). Level (million pesetas) Annual growth rates (%) Region 1955 1980 2005/2010 1955 to 1980 1980 to 2005/2010 1955 to 2005/2010 AN Andalusia 84,854 224,431 363,687 4.0% 1.8% 2.8% AR Aragon 22,814 64,116 93,543 4.2% 1.4% 2.8% AS Asturias 8,307 9,472 6,541 0.5% -1.4% -0.5% CB Cantabria 2,141 2,887 1,880 1.2% -1.6% -0.2% CL Castile and Leon 79,737 132,993 114,888 2.1% -0.5% 0.7% CM Castile-La Mancha 44,508 119,454 107,900 4.0% -0.4% 1.7% CT Catalonia 31,577 87,691 77,362 4.2% -0.5% 1.7% EX Extremadura 22,137 53,936 93,546 3.6% 2.1% 2.8% GA Galicia 34,348 56,970 54,056 2.0% -0.2% 0.9% IB Balearic Islands 5,929 13,097 10,791 3.2% -0.7% 1.2% MC Murcia 7,836 32,340 72,200 5.8% 3.0% 4.4% MD Madrid 6,592 9,640 5,783 1.5% -1.9% -0.3% NC Navarra 10,855 22,694 26,713 3.0% 0.6% 1.7% PV Basque Country 6,917 8,722 7,163 0.9% -0.7% 0.1% RI La Rioja 6,821 16,143 15,035 3.5% -0.3% 1.5% VC Valencian Comm. 44,785 101,584 89,958 3.3% -0.4% 1.4% TOT Total 420,158 956,170 1,141,047 3.3% 0.7% 1.9% 6Miguel Martín-Retortillo, Ana Serrano and Ignacio Cazcarro Spanish Journal of Agricultural Research September 2020 • Volume 18 • Issue 3 • e0107 olive oil and wine (the latter up until 1980). Both citrus and non-citrus fruits moved from a share of about 4% to about 7–8% at the end of the period11 . All of these products are high value-added crops. Besides, fodder crops also increased in importance in crop production, moving from a share of 6% to 11%. We have then clearly seen this double concentration: reallocation to the south and east of Spain, and specialization in high value-added products. This came about thanks to several factors. First, Spanish exports of vegetables and fruit rose dramatically to become the leading agricultural export products in recent years; they are highly profitable and are the clearest representation of irrigated production (particularly in the southeast of Spain). This is not a new specialization pattern: Spain has traditionally exported fruit and vegetables ever since the last decades of the nineteenth century, when its exports represented approximately one third of the global exports of Mediterranean horticultural products. During the second globalization, the diversification of agricultural exports and the greater importance of transformed products in international trade did not change the preponderance that fruit and vegetables have enjoyed in the agricultural sector. The relative and absolute increase in the production of vegetables, fruit and oil was also driven by internal demand. The rising Spanish population necessitated higher production to meet the food needs. Besides, economic development 11 In the case of citrus trees, Valencia is one of the main producers of these products, despite small size of these farms (Reig & Picazo, 2004). triggered improvements in the Spanish diet, which moved towards greater consumption of fruit, vegetables and oil. In the case of wine, internal Spanish demand fell abruptly in the second period, despite its increasing export orientation. Thus, Spanish wine production flowed increasingly to the international markets. These decreases in production in a few products go to explain the figure in Table 2 for total production in the last period, which did increase compared to the 1980s, but at a more modest rate than before. The changing diet of Spaniards also involved greater meat consumption. Moreover, there was a big increase in meat exports, especially after the 1980s. This was reflected in the greater importance of livestock production during the second globalization, which at the same time required larger volumes of animal feed, both domestically produced and imported. During the early decades of the second half of the twentieth century, there was also a marked increase in the production of non-wheat cereals for animal feed, such as barley and maize. As illustration, at the beginning of the 1960s, 39% of cereal production was for animal feed, but by the beginning of the 1980s this figure had reached 72%. Nevertheless, as Table 2 shows, over the period as a whole there was a big decline in the share of cereals – from 47% in 1955 to 14% in 2005/2010. The decline in some low-value products (such as legumes, and other woody crops) was also striking; this Figure 1. Monetary value of production in 1955 (size of the circles), absolute changes between 1955 and 2005/2010 (colour intensity of the circles) and relative (%) change between 1955 and 2005/2010 (colour of the provinces). Table S1 [suppl.] for the names of the provinces and the relation to wider administrative regions. Source: Own elaboration from MAPAMA (1955, 1980, 2005, 2010). Spanish Journal of Agricultural Research September 2020 • Volume 18 • Issue 3 • e0107 7 Double concentration explaining the outstanding increase in Spanish crop production was mainly due to a fall in their consumption, especially after 1980. Causes of the double concentration: decomposition results In this section, we quantify and expand on the causes of the double concentration seen in previous sections. To that end, first we examined the value decomposition, which analyses changes in the volumes and prices as drivers of monetary production. Then we looked at the determinants that triggered the increases in crop production. As indicated in the methodology, the reason for the existence of a price effect is that, although the effect of prices is removed at the aggregate level of whole crop production, the individual prices of products evolved in different ways. These differences may show those crops that experienced higher increases in their value added, which thus encouraged their production and generated additional income. In Fig. 2a we found a negative price effect for cereals from 1955 to 1980, given that cereal price fell or remained stable. However, we observe a positive quantity effect 12 Nevertheless, the aridity of these provinces generates problems of availability of resources, especially water (Cazcarro et al., 2015b). (i.e. there was an increase in physical units of production), which contributed to a net increase in the value of production. In regional terms, this happened in Castile-La Mancha and notably Castile and Leon, two areas with significant cereal production. Despite the generally stronger impact of the quantity effect, the price effect explained more than 33% of the change, particularly for olive oil and vegetables. Andalusia accounts for an enormous share of these groups of products. In the case of olive oil, the evolution of its production and export was slightly erratic between 1980 and the first years of the twenty-first century. As for vegetables, during the last decades of our period of study, their production kept increasing via quantity effects, and price effects showed relatively greater importance. These changes were geographically concentrated in Andalusia, Extremadura and Murcia (Extremadura is more in the west, but there was an important concentration in the south and east of Spain), areas with optimal conditions for these crops: a high number of sunshine hours per year and water resources from the expansion of irrigation.12 Furthermore, increased external demand for this product type generated growing exports. The positive price effect was a strong incentive to produce more Table 2. Monetary production value (prices deflated using the base year 1980) by product group (in million pesetas) and annual growth rates (%) [1] Oil refers to olive oil, olive consumption are olives that are directly consumed, wine is considered as an individual item. Grapes and raisins for consumption are included as non-citrus fruits. Source: Own elaboration from Spain’s Agrarian Statistics Yearbooks for 1955, 1980 and 2005 to 2010 (MAPAMA, 1955, 1980, 2005, 2010). Product group[1] Monetary production value Annual growth rate (%) 1955 1980 2005/2010 19551980 19802005/10 19552005/10 Oil 22,143 55,483 127,276 3.74 3.12 3.42 Cereals 195,687 270,878 156,887 1.31 -2.00 -0.42 Industrial crops 25,422 69,295 33,185 4.09 -2.69 0.51 Fodder 26,447 85,907 124,578 4.83 1.39 3.03 Citrus fruit trees 16,011 46,039 84,135 4.32 2.26 3.24 Non-citrus fruit trees 16,602 73,739 95,732 6.15 0.97 3.43 Nuts 10,276 23,964 42,915 3.44 2.18 2.79 Vegetables 43,739 170,987 363,843 5.60 2.84 4.16 Legumes 21,836 16,488 7,538 -1.12 -2.86 -2.02 Olive consumption 1,954 5,372 13,851 4.13 3.57 3.84 Other woody crops 4,444 2,418 1,114 -2.41 -2.83 -2.63 Tubers 32,878 58,854 30,504 2.36 -2.40 -0.14 Wine 2,719 76,747 59,489 14.29 -0.94 6.11 Total 420,158 956,170 1,141,047 3.34 0.66 1.94 8Miguel Martín-Retortillo, Ana Serrano and Ignacio Cazcarro Spanish Journal of Agricultural Research September 2020 • Volume 18 • Issue 3 • e0107 vegetables, substituting low value-added for high value-added products, which reveals concentration in terms of products. The case of fodder crops is also striking. Despite a fall in the quantity of production, prices boosted the value of production. This happened in northern regions, such as Asturias, Cantabria, Vizcaya, Guipúzcoa and other northern provinces (see Fig. S1a [suppl.]) with high rainfall and high yields. Finally, the value of wine production fell since the 1980s – because of both the quantity and price effects, although the quantity effect was positive from 1955 to 1980. This decrease in monetary wine production from 1980 (Table 2 and Fig. 2a) hinders large regional heterogeneities. According to the information from the Spain’s Agrarian Statistics Yearbooks for 1980 and 2005 to 2010 (MAPAMA, 1955, 1980, 2005 & 2010) wine production notably increased in the north/northeast of the country (regions of La Rioja, Navarra, Barcelona and Huesca). However, between 1980 and 2005/2010 important decreases happened in Castilla-La Mancha (for example in the province of Ciudad Real), driving the results at the national level. Cañizares Ruiz & Ruiz Pulpón (2014) explained that the 13 To evaluate the robustness of these results we calculated the value of crop production in 1955, 1980 and 2005/2010 at constant 1980 prices and applied the same decomposition as in Table 3. The decomposition of the value of crop production using deflated prices (Table 3) is very similar to that using 1980 constant prices. Concretely, for the period 19552005/2010 the differences between both approaches range around 1% and 2,5% for the different effects. Thus, the decomposition is robust to the selection of a constant or deflated base year. 14 This number obviously matches the differences shown in Tables 1 and 2 of value in 2005/2010 of 1,141,047 million pesetas and in 1955 of 420,158 million pesetas. EC regulation entailed the loss of approx. 130,000 ha of vineyards in Castilla-La Mancha from 1987 to 1999 (a fall of around 20% of the existing area). Despite that at the same time there was a considerable expansion of irrigated surface, it was not enough to offset the huge fall of rain-fed area. The former results can be complemented with the quantification of the changes in regional and crop composition as determinants of the change in Spanish crop production. In Table 313 we observe that the Spanish value of crop production increased from 1955 to 2005/2010.14 The findings from LMDI decomposition indicate that this growth was essentially (84%) explained by the SE. The CE (16%) also goes some way to explaining the change in the value of production; while the share effect was almost non-existent across this long period. Differentiating between the first and the second period studied, we observe how the scale effect was larger in the second period, while the composition effect was larger in the first. Although the figures are small, it is interesting that regional changes led to an increase in the value of monetary production in the first period, but to a decrease in the second. Figure 2. Decomposition of the changes (in million pesetas) into the price and quantity effects (a) and into the composition, share and scale effects by region (b). AN: Andalusia; AR: Aragon; CL: Castile and Leon; CM: Castile-La Mancha; CT: Catalonia; EX: Extremadura; GA: Galicia; MC: Murcia; and VC: Valencian Community. ‘Other’ comprises small regions (AS: Asturias; CB: Cantabria; IB: Balearic Islands; MD: Madrid; NC: Navarra; PV: Basque Country; and RI: La Rioja) and their graph is shown in Table S1 [suppl.]. Source: Own elaboration (MAPAMA, 1955, 1980, 2005, 2010). Spanish Journal of Agricultural Research September 2020 • Volume 18 • Issue 3 • e0107 9 Double concentration explaining the outstanding increase in Spanish crop production We saw in Fig. 2a that monetary production increased in most regions; in Fig. 2b, we see that this is largely explained by scale effects, but that the positive composition effects also play a part. Production tended to move to more valued products or high value-added products. As in Fig. 2a, the case of Andalusia is very illustrative. This region led the growth in value. We can see that although for this region there are purely scale effects (most regions increase their monetary production), there are also notable regional/share effects (production moved to this region) and product composition effects (higher specialization in more valued products). In this regard, the spatial concentration had a negative share effect in terms of production for regions such as Castile and Leon, Castile-La Mancha, Valencian Community, Cantabria, and many others such as Galicia, Madrid, the Basque country,15 etc. This means that these regions reduced the production of some crops, not because those crops were less in evidence in Spain’s production as a whole, but because production moved elsewhere. As hinted above, this ‘elsewhere’ included Andalusia, Extremadura, Murcia and Aragon. Although globally share effects seem to have had a small influence on changes in production, they have important explanatory power when we look at the regional detail. Discussion In this work, we have analysed the structural changes in agricultural production in Spain during the second half of the twentieth century and the beginning of the twenty-first, in the framework of major transformations in the agricultural sector itself, as well as in the institutional and political context. We assessed how changes in the volume and price of production have conditioned the value of crop production in the long term. Then, we examined changes in crop specialization and regional composition as drivers of monetary crop production. We found strong evidence for the existence of a ‘double concentration’, in space and crops. 15 See for the small regions (AS: Asturias; CB: Cantabria; IB: Balearic Islands; MD: Madrid; NC: Navarra; PV: Basque Country; and RI: La Rioja) Fig. S1b [suppl.] 16 Several differences are found between the type of the irrigation (Rodríguez-Díaz et al., 2008). The highly detailed regional and product disaggregation is an important value added of this paper, as it allows in-depth study of the long-term changes experienced in the Spanish agricultural sector. In terms of methodology, after completing the database and deriving the main descriptive trajectories for agricultural production, we used index decomposition analysis (IDA). This technique is useful in providing insights into the factors that lie behind the structural changes experienced by the Spanish agricultural sector. Our article includes empirical evidence on the factors that enabled Spanish agricultural production to grow so conspicuously in European terms. We observe that the heterogeneous growth in Spanish crop production was based on a double concentration: agriculture focused on high value-added products, such as vegetables, fruit and olive oil; and, spatially it was concentrated in the south and east of Spain. The expansion of irrigation, the evolution of internal demand and the progressive integration into international markets all go to explain this double concentration. In particular, crop production increased at 2% yearly from 1955 to 2005/2010; this growth was particularly intensive between 1955 and 1980, a period when the annual growth rate stood at 3.3%. Thus, production soared (as revealed by the scale effect) and entailed a spatial concentration, notably in the south and east of Spain (as shown by the share effect). Andalusia showed its growing importance in agricultural production, with a share that went from 20.2% of the value of production in 1955 to 31.9% at the beginning of the twenty-first century. The relative increase mostly occurred from 1980. This region bolstered the tradition of cultivating products for export and products with high value-added (Serrano & Pinilla, 2014; Hernández et al., 2016; Aparicio et al., 2018). The expansion of irrigation facilitated increased vegetable production (6% annually throughout the period studied) and olive production (3.6% annually)16 . Other regions like Extremadura, Murcia or Aragon also specialized in products that enjoyed strong demand, acquiring a greater share of crop production. Table 3. Logarithmic Mean Divisia Index (LMDI-I) multiplicative decomposition of value of crop production changes (million pesetas) CE: composition effect. RE: share effect. SE: scale effect. Source: Own elaboration from MAPAMA (1955, 1980, 2005, 2010). CE (%) RE (%) SE (%) Prod 1955 to 1980 87,016 16 861 0.2 448,136 84 536,012 1980 to 2005/2010 25,136 14 -1064 -0.6 160,804 87 184,877 1955 to 2005/2010 112,152 16 -203 0.0 608,940 84 720,889