Socioeconomía

 

Determinants of labour productivity convergence in the european agricultural sector

 

Determinantes de la convergencia en productividad del trabajo en el sector agrario europeo

 

María Carmen Cuerva*

 

Facultad de Ciencias Económicas y Empresariales. Universidad de Castilla–La Mancha. Plaza de la Universidad, 2, 02071. Albacete, España. * Author for correspondence (mariac.cuerva@uclm.es).

 

Received: June, 2011.
Approved: August, 2012.

 

Abstract

The EU agriculture, in spite of the market integration process, shows important differences in terms of productivity and efficiency. It is thus important to determine whether there is a convergence process in agricultural productivity and what factors would explain the observed disparities. To respond to these two questions, the dynamic of the agricultural labour productivity of 125 EU regions was analysed. The methodology was an estimation of the convergence equation with cross–section data for the period 1985–2004. The results showed that there is a very slow process of absolute convergence in productivity among the EU agricultures. Besides, productivity growth is related to agricultural out–migration and higher levels of physical capital. On the contrary, Common Agricultural Policy (CAP) support to the agricultural sector does not seem to have contributed to the productivity growth.

Key words: European regions, labour agricultural productivity, beta convergence, Common Agricultural Policy.

 

Resumen

La agricultura en la UE, a pesar del proceso de integración del mercado, muestra importantes diferencias en términos de productividad y eficiencia. Por tanto, es importante determinar si hay un proceso de convergencia en la productividad agrícola y los factores que explicarían las diferencias observadas. Para responder a estas dos preguntas se analizó la dinámica de la productividad del trabajo agrícola de 125 regiones de la UE. La metodología empleada fue una estimación de la ecuación de convergencia con datos de corte transversal para el período 1985–2004. Los resultados mostraron que hay un proceso muy lento de convergencia absoluta en la productividad entre las agriculturas de la UE. Además, el crecimiento de la productividad está relacionado con la emigración agrícola y con mayores niveles de capital físico. Por el contrario, el apoyo de la Política Agrícola Común (PAC) al sector agrícola no parece haber contribuido al crecimiento de la productividad.

Palabras clave: regiones europeas, la productividad del trabajo agrícola, convergencia beta, política agrícola común.

 

INTRODUCTION

The final reason of the economic integration in the EU is to reach the economic and social cohesion (convergence) among territories. In this sense, the European Commission (2008) considers that agricultural sector has a relevant role in relation to the objective of cohesion; in other words, it is a priority area of convergence (Sassi, 2010). Therefore, the reduction of disparities within the sector has been a great concern for economic policy makers since the beginning of the European integration process.

From the perspective of the neoclassical theory of economics integration, the reinforcement of the European market integration should imply convergence in terms of productivity and efficiency levels of the different sector activities (Sala–i–Martin, 1996; Gardiner et al., 2004). Nevertheless, in spite of this market integration, the EU agricultural sector is very heterogeneous. This implies productivity and income differences among the European agricultures. Differences in levels of employment, investment, human capital and public support through the Common Agricultural Policy (CAP) highlight important disparities in the European agriculture. These agriculture–specific related characteristics condition the possibilities of endogenous development of territories given that each region differs in the endowment of these factors affecting its results in terms of productivity growth. In this sense, the expected convergence activated with the integration process should not be automatic. On the contrary, the convergence–or divergence–generating mechanisms are more extended and complex than the ones stemming from neoclassical approaches. That means that mechanisms which explain the convergence process under the neoclassical assumptions (the diminishing returns of capital, free factor mobility, free trade and technological progress diffusion) could be counteracted by the differences in the allocation of human, technological and public capital, knowledge or investment levels which condition the convergence process and entail that only regions with common characteristics converge to similar stationary states of productivity^[1].

Interest in regional convergence studies have recently increased according to the expansion and deepening of the European integration process. Many studies have tried to provide evidence from the EU convergence process in terms of regional GDP per capita or per worker, but they do not pay much attention to sector considerations and, in particular, to agricultural sector (Alexiadis and Alexandrakis, 2008).

Therefore, few studies have tested for regional agricultural output and labour productivity convergence. There are some relevant studies, but evidence on this topic is still limited. Paci (1997) and Paci and Pigliaru (1999) take sector into account and they find no evidence of absolute convergence in labour productivity in a sample of 109 European agricultures during the eighties. Convergence only occurs inside groups of similar agricultures and it has been quicker in the Northern regions of Europe.

Other studies go further and deal with the agricultural regional convergence in detail. Colino et al. (1999) and Colino and Noguera (2000) analyse the intra–sector convergence in 98 European agricultures in terms of labour productivity and show the inexistence of absolute convergence because of the performance of the most productive regions. Convergence only occurs among agricultures with similar structural and productive characteristics; but in contrast with the above–mentioned papers, convergence is higher among Southern regions.

Castillo and Cuerva (2005) confirm that agricultures located in less developed regions converge more quickly to their stationary state level of productivity than the more developed ones. These authors study the differences in the process of convergence taking sector specialization into account and conclude that regions specializing in Continental products, with great public support from the CAP compared with Mediterranean products, have got closer to their stationary state earlier. This result leads the authors to recognize the positive effect of the CAP support on the convergence process in Continental regions.

The most recent studies also suggest that European agricultures follow a pattern of club–convergence and they do not converge to the same level of productivity. This pattern is attributed to differences in the farm's size (Alexiadis and Alexandrakis, 2008), in technological capital accumulation (Sassi, 2010) or in levels of regional development and in the sector investment (Ezcurra et al. 2008, 2011).

Therefore, the aim of this paper was: 1) to test the convergence hypothesis under the assumption that convergence does not occur at the same productivity stationary state, that is convergence is not absolute and does not occur automatically with the integration of the market; 2) to identify the factors which could foster productivity to converge. In this sense we consider movements of labour force, endowment of human capital, investment flows and public capital (measured by the CAP market support). Among these factors, the main attention will be addressed to the role of the CAP market support. Is the European agricultural policy contributing to the sector convergence and to increase productivity? Are the production subsidies the best way to foster convergence within the sector?

 

MATERIAL AND METHODS

The most common methodology to measure convergence consists of regressing time–averaged productivity growth on its initial level for a cross–section of economies. This model is known as convergence equation, developed by Barro and Sala–i–Martín (1990) according to Solow's growth equation. The form is:

where y_i is the agricultural productivity of the region i (i=1,2,...N), a isa constant which reflects the variables determining the stationary state, u is a disturbance term which represents the unexpected changes in the production conditions or preferences (with zero mean and constant variance which is distributed independently of the explanatory variable), T is the length of the considered period, 0 refers to the initial year, and t is the final year.

The correlation between the initial productivity level and its rate of growth must be negative and statistically significant. That will mean the backward regions grow more than the most productive ones, so that they approach a common and unique level of productivity in the stationary state. This process is known as absolute β–convergence.

The main advantage ofequation (1) is that it could be estimated by Ordinary Least Squares (OLS), so b = — (1 / T)(1 — e^–βT) where the β coefficient represents directly the speed at which regions converge to the stationary state.

This kind of convergence occurs among regions which only differ in the initial level of productivity and have the same factors determining the stationary state, represented by the coefficient a. However, if a is not constant across regions, the model will be wrongly specified. This fact could seriously affect the robustness of the convergence coefficient and lead to misleading results and conclusions. To avoid this problem appropriate variables which identify the differences in the stationary states were added to the right side of equation (1):

Π is a structural parameter attached to X_i,t , which represents a set of variables included in the equation with the aim of controlling the differences in the stationary states. If b has the correct sign and one of the variables is significant, then conditional β–convergence will be identified.

The difference between equations (1) and (2) is that the stationary states could differ among regions. In this sense, the level of productivity in each stationary state derived from (2) would be:

The long–term equilibrium level depends on a set of structural variables which can vary across the diverse economic areas.

The estimation of convergence equations was done with STATA 10.

Data issues

Productivity is measured as the real GVA (Gross Value Added) per worker at basic prices. Data on agricultural GVA and employment are provided by Cambridge Econometrics European regional database for the period 1985–2004, which is designed to cover all EU regions and makes comparative analyses possible. Additionally, it completes the lacking information in Eurostat regional database for several sector– and regional–level variables, such as production, employment or investment, resorting to official national statistics.

When it comes to selecting the territorial unit, it is important that the largest regions are not overvalued in the set of data used. This could happen if we limit the information to the NUTS–2 level^[2]. To get a greater homogeneity a combination of the different NUTS levels it has been used. The sample has 125 EU–15 territorial units (Table 1).

Figure 1 shows the regional agricultural productivity relative to the EU–15 mean in 1985 and 2004. There are considerable differences across regions. For example, in 1985 the productivity of most Southern regions of Europe belonging to Portugal, Spain, Italy and Greece was below 75 % of the European average, while a large number of regions belonging to Central and Northern countries such as France, Denmark, United Kingdom or Sweden maintain productivity levels above 150 % of the mean value. In 2004, this picture had marginally changed and disparities were still evident.

To identify the factors which could explain these observed differences, several explanatory variables related to sector productivity were selected (Table 2). A correlation analysis was made among explanatory variables and the lowest correlation coefficient was —0.038 (non significant).

The accumulation of physical capital is a necessary condition for sustained productivity growth (Gutierrez, 2000). If technology is incorporated in capital goods, their increase could speed the introduction of technological progress and improve sector efficiency. To capture this effect, sector–specific investment, measured as the real Gross Capital Formation (GCF) per worker at the initial year, was included in the specification^[3].

Lockheed et al. (1980) and Lassibille (1986) underline the importance of human capital as a productivity growth source Human capital improves farmers' management skills and enhances efficiency in the use of the productive factors. Human capital also improves the performance of physical and technological capital so that, indirectly, it influences productivity. Both effects increase productivity and affect the stationary state. From the year 2000 onwards, the Farm Structure Survey (Eurostat) publishes the percentage of farmers with basic and full education attained at the European regional level^[4]. It does, therefore, not provide information for the beginning of the period. With the available data, the percentage for the year 2000 obtained by the EU Farm Structure Survey 2000 was introduced into the model as a mean value of the period^[5].

Agricultural sector migration is another conditioning variable on productivity. Although some studies examine the role of the out–migration in the sector (Gutierrez, 2000; McErlean and Wu, 2003), the impact of this migration on the rate of productivity growth in the agricultural sector is still an open question. The loss of agricultural employment may lead to considerable productivity increases, in the case that labour force was inefficient. In this scenario, out–migration raises the rate of agricultural productivity convergence (McErlean and Wu, 2003). To take this into account, following the methodology of Larson and Mundlak (1997) and Gutierrez (2000), the sector–migration rate towards the remaining productive sectors was calculated. Without migrations, agricultural and non–agricultural employment had grown at the same rate of overall employment. Deviations from this rate are due to migration, so that the total number of emigrants from the agricultural sector can be calculated as follows:

where n is the growth rate of the total labour force at the regional level during the whole period, and L₀ and L_t are the level of agricultural employment in the initial and final year.

Dividing M_t by L₀ and calculating the mean ratio for the period 1985–2004, the mean migration rate is obtained.

Finally, the role of the CAP is taken into consideration as a potential conditioning factor in productivity. It is important to find out whether its intervention mechanisms have provided an incentive for agricultures to reinforce competitiveness and efficiency and to reduce disparities. Given the lack of complete official data, CAP–related data was taken from the Study on the Impact of Community Agricultural Policies on Economic and Social Cohesion (European Commission, 2001) which contains information about direct payments and market price support at regional level for 1989, 1994 and 1996. There is no information for the initial year of the period. For that reason, data for 1989, the closest year to the initial one, is used^[6]. In this case information for the regions of Austria, Sweden and Finland is lost because of their adhesion to the EU in 1995. The weight these payments represent in the sector production was introduced in the convergence equation. Data are expressed in constant prices using the national Gross Domestic Product (GDP) deflator.

 

RESULTS AND DISCUSSION

Table 3 illustrates the results of testing agricultural absolute convergence with cross–sectional data according to equation (1). The OLS regression was estimated using White's correction for heteroskedasticity to get consistent standard errors. The coefficient of initial productivity is negative and significant. The annual speed of convergence is 0.91 %. This evidence lends support to the conclusion that the process of convergence is very slow. In 2004 the European agricultures were still very far away from a hypothetical common stationary state. These results are in line with the papers of Gil Canaleta (2001) and Castillo and Cuerva (2005), but in our study the speed of convergence is lower.

To test whether or not the convergence process has been homogeneous the period of time has been broken down into two homogeneous intervals, 19851994 and 1994–2004. The Wald test was used to prove if β is constant throughout time and it showed that the β estimation is stable between sub–periods. In 1985–1994, the annual speed of convergence was 1.50 %. In 1994–2004 the speed of the process lost significance and decreased up to the rate of 0.85 %. These results confirm that convergence has weakened since the second half of the nineties.

Nevertheless, the explanatory power of the estimations is low (low model fit), which points to large unexplained regional productivity differentials. In this case, the β parameter could be underestimated and the possibility of converging to different stationary states (not absolute convergence) should be tested.

A conditional convergence model is estimated in the pursuit of identifying the differences across the stationary states. Table 4 illustrates the results of estimations for the different periods according to equation (2). In 1985–2004 and in 1985–1994 estimations only include the EU–12 regions. In 19942004 the estimation includes the EU–15 regions because of the availability of CAP data for the new Member States in 1995.

Statistically significant β–convergence is observed across the whole period. The value of the parameter involves an annual speed of convergence of around 1.6 %.

The sector investment and the migration rate are the only significant variables registering the expected sign in the period 1985–2004. Ceteris paribus, an investment improvement increases the rate of productivity growth. Gutierrez (2000) reached the same conclusion. Agricultures with low levels of investments have a great potential to develop in this way and future increases in their capital gross formation will mean higher productivity growth. Nevertheless, the explanatory power of this variable is low (0.003 in 1985–2004). That means that an increase in investment per worker of 10 % raises the productivity growth rate by 0.003 percentage points per years. This result with the fact that investment is only significant in the whole period could mean that its effect on productivity takes place in the long term. Out–migration appears to be a major element in the productivity growth (the estimated coefficient is 0.5 in 1985–2004). This result is not surprising because the loss of the sector labour force is in part due to the mechanization which increases sector productivity. Additionally, the migrations of the workers together with the greater skills of the remaining workers cause favourable effects on growth productivity in this period. De la Fuente and Freire (2000), Gutierrez (2000) and McErlean and Wu (2003) find similar evidence. This result may support the idea that migration from agriculture to the rest of economic sectors has positively influenced the intra–sector convergence. However, this significant effect seems to have disappeared in the recent period.

In relation to human capital, the results differ from what is expected. Although theoretical and technical knowledge has been essential to reduce disparities and to improve food security (FAO, 2000) it does not seem to have a significant effect on productivity growth. However, in 1994–2004 certain statistical significance is observed but the estimated coefficient is literally zero. Alfranca (1998) and Serrano (1999) reached a similar conclusion. One possible explanation could be that the indicator used is not suitable because it does not refer to the initial period. In addition, human capital measurement is complex. Different perspectives of knowledge must be taken into account: education, labour experience, learning by doing or workers' skills. The selected explanatory variable does not consider all those aspects. In many studies, the results about the relation between education and productivity are disappointing. To a large extent, it is due to the indicators used (De la Fuente, 2005).

With regard to the CAP, its coefficient is not significant. According to Sassi (2010), this result calls into question the role of the CAP as an important factor in increasing productive efficiency. From the side of a support market policy, all the agricultural sectors have not strongly benefited from the CAP (Vega, 2005; García álvarez–Coque and Wieck, 2001). In this sense, the CAP could have interfered in productive specialisation, in favour of the most supported productions, and may have impeded the exploitation of the competitive advantages in each agriculture system. According to Esposti (2007), the effect of CAP payments is exerted by maintaining more sector labour force than required, making productivity gains difficult.

As a result of the 1992 CAP reform, direct payments as mechanisms of market support were introduced. This concern may explain why the change seems not to have a positive effect on productivity growth in the period 1994–2004. Direct payments may have stimulated small farmers to remain as formal farm holders in order to receive these payments even though they are often not fit enough to farm the land because of their advanced age, being employed in other sectors or even living in urban areas. This could help to explain why migration rate is not a decisive factor to explain productivity growth any longer in the period 1994–2004.

These conclusions need to be taken cautiously. The non–significance of the CAP's first pillar support could also indicate that its effect on intra–sector convergence has been too small to be detected. In fact, the CAP support is relatively small compared to other financial sources and only a few farms receive the main support: in the eighties, 80 % of support ended up in only 20 % of farms (European Commission, 1991). Moreover, its lack of significance could be explained by potential mis–specification of the CAP information in this period. Anyway, the non–significance of the CAP may not be interpreted as lack of effectiveness.

 

CONCLUSIONS

The analysis of the evolution of agricultural labour productivity for European regions concludes that the market integration process has not contributed to the absolute convergence among European agricultures. The agriculture–specific related characteristics have limited the convergence process. From the economic policy point of view, the existence of a common agricultural policy (CAP) should foster the convergence process. Nevertheless, the evidence seems to indicate that it is not so.

This conclusion does not should be taken as CAP is inefficient (we have to remind that market support has allowed the survival of a lot of small holdings and farmers in Europe, very important for the conservation of the rural areas in the EU) but it should make think policy makers about alternative measures less distorting of market and international trade to improve efficiency of European agriculture. The solution is not to increase the market support. It is more urgent to apply a sector policy capable of generating employment, diversifying activities and reducing the structural problems of holdings. In this sense, fostering productive and R&D investment should be in the core of the CAP measures instead of production subsidies.

 

LITERATURE CITED

Alexiadis, S., and A. Alexandrakis. 2008. Threshold Conditions' and Regional Convergence in European Agriculture. International J. Econ. Sci. Appl. Res. 1: 13–37.         [ Links ]

Alfranca, O. 1998. Determinantes de la productividad total de los factores en el sector agrario español. Investigación Agraria. Producción y Protección Vegetales 13: 201–220.         [ Links ]

Barro, R., and X. Sala–i–Martín. 1990. Economic growth and convergence across the United States. NBER Working paper 3419. Cambridge. pp: 61.         [ Links ]

Castillo, J. S., and M. C. Cuerva. 2005. Análisis de la convergencia en productividad agraria en las regiones europeas. Economía Agraria y Recursos Naturales 10: 89–106         [ Links ]

Colino, J., P., Noguera, and M. Rodríguez. 1999. La agricultura gallega en la Unión Europea: Balance del primer decenio. Fundación Caixa Galicia. 214 p.         [ Links ]

Colino J., and P. Noguera. 2000. Patrones estructurales y convergencia interregional en la agricultura europea. Historia Agraria 22: 111–128.         [ Links ]

De La Fuente, A. 2005. Educación y crecimiento: un panorama. Working Paper D–2005–03, Ministry of Economics. Spain. pp: 42.         [ Links ]

De La Fuente, A., and M. J. Freire 2000. Estructura sectorial y convergencia regional. Revista de Economía Aplicada 23:189–205.         [ Links ]

Esposti, R. 2007. Regional growth and policies in the European Union: Does the Common Agricultural Policy have a counter–treatment effect? Am. J. Agric. Econ. 89: 116–134.         [ Links ]

European Commission. 1991. The development and future of the CAP. Reflection paper of the Commission. COM (91) 100 final. Brussels. pp: 21.         [ Links ]

European Commission. 2001. Study on the impact of community agricultural policies on economic and social cohesion. Directorate General for Regional Policy. Brussels. pp: 108.         [ Links ]

European Commission. 2008. Green Paper on Territorial Cohesion Turning territorial diversity into strength. COM (2008) 616 final. Brussels. pp: 12.         [ Links ]

Ezcurra, R., B. Iraizoz, P. Pascual, and M. Rapún. 2008. Spatial disparities in European agriculture: a regional analysis. Appl. Econ. 40: 1669–1684.         [ Links ]

Ezcurra, R., B. Iraizoz, P. Pascual, and M. Rapún. 2011. Agricultural productivity in the European Regions: Trends and explanatory factors. Eur. Urban and Regional Studies 18: 113–135.         [ Links ]

FAO (Food and Agriculture Organization). 2000. The State of Food and Agriculture. FAO Collection. Agriculture 32. pp: 329.         [ Links ]

García Alvarez–Coque, J. M., and C. Wieck. 2001.¿Es la reducción de las desigualdades interregionales una cualidad de la PAC? 4th National Conference of Agricultural Economics Association. Pamplona. Spain. pp: 23.         [ Links ]

Gardiner, B., R. Martin, and P. Tyler. 2004. Competitiveness, productivity and economic growth across the European regions. Regional Studies 38: 1037– 1059        [ Links ]

Gil Canaleta, C. 2001. Capital Público y Convergencia en las Regiones Europeas. Civitas, Madrid. 244 p.         [ Links ]

Gutierrez, L. 2000. Convergence in US and EU agriculture. Eur. Rev. Agric. Econ. 27: 187–206.         [ Links ]

Larson, Y., and D. F. Mundlak. 1997. On the intersectorial migration of agricultural labour. Econ. Develop. Cultural Change 45: 295–319.         [ Links ]

Lassibille, G. 1986. El papel del capital humano en la agricultura española. Agricultura y Sociedad 40: 37–66.         [ Links ]

Lockheed, M. E., D. T. Jamison, and L. J. Lau. 1980. Farmer education and farm efficiency: a Survey. Econ. Develop. Cultural Change 29: 37–76.         [ Links ]

McErlean, S., and Z. Wu. 2003. Regional agricultural labour productivity convergence in China. Food Policy 28: 237–252.         [ Links ]

Paci, R. 1997. More similar and less equal. Economic growth in the European regions. Weltwirtschaftliches Archiv. 133: 609–634.         [ Links ]

Paci, R., and F. Pigliaru. 1999. European regional growth: Do sectors matter? In: Adams, J., and F. Pigliaru (eds). Economic Growth and Change. National and Regional Patterns. Edward Elgar. Cheltenham. pp: 213–235.         [ Links ]

Sala–i–Martin, X. 1996. Regional cohesion: evidence and theories of regional growth and convergence. Eur. Econ. Rev. 40: 1325–1352.         [ Links ]

Sassi, M. 2010. OLS and GWR Approaches to agricultural convergence in the EU–15. Int. Adv. Econ. Res. 16: 96–108.         [ Links ]

Serrano, L. 1999. Capital humano, estructura sectorial y crecimiento en las regiones españolas. Investigaciones Econ. 23: 225–249.         [ Links ]

Vega, J. 2005. ¿Quiénes son los verdaderos beneficiarios de la PAC? In: García Delgado, J. L., y J. García Grande (eds). Política Agraria Común: Balance y Perspectivas. Colección Estudios Económicos 34. Servicios de Estudios de La Caixa. pp: 100–123.         [ Links ]

 

NOTES

¹ The Market Integration theory also assumes equal conditions for financing and taxes to converge. Farmers with a lower financial debt or taxes adjust more easily to market conditions and thus tend to be more efficient and productive. Therefore, different situations in financial terms may imply different levels of productivity growth of holdings and may delay the convergence process. Nevertheless, an analysis of the impact of the financial and tax conditions on productivity convergence could be more appropriate in a microeconomic context, where the unit of analysis were the holding but not the region.

² The Nomenclature of Territorial Units for Statistics (NUTS) was created by Eurostat with the purpose of making a unique and uniform breakdown of territorial units of the EU. The NUTS is a hierarchical classification where each Member State is divided into a number of NUTS–1 regions; each one of them, at the same time, is divided into a number of NUTS–2 regions and, at the same time, and then divided into NUTS–3 regions.

³ It could be more interesting to analyze the effect on convergence of the Net Capital Formation. The lack of data for this variable impedes to do this empirical exercise.

⁴ According to the European Commission this indicator is defined as follows: 1) Only practical agricultural experience: experience acquired through practical work on an agricultural holding. 2) Basic agricultural training: any training courses completed at a general agricultural college and/or an institution specialising in certain subjects such as horticulture, viticulture, sylviculture, among others. A completed agricultural apprenticeship is regarded as basic training. 3) Full agricultural training: any training course continuing for the equivalent of at least two years full time training after the end of compulsory education and completed at an agricultural college, university or other institute of higher education in agriculture, horticulture, viticulture, sylviculture, agricultural technology or an associated subject.

⁵ Data are not available for Sweden and the Finnish regions of Saarland and Aland.

⁶ Data are not available for the English regions of the North East, North West and South East for 1989 and 1994. In 1996 there are data for Austria, Sweden and Finland, but the information is not available in the aforementioned English regions and in three Swedish regions.