WHAT IS BEHIND BIASED TECHNICAL CHANGE IN PRODUCTION OF CEREAL AND OILSEED CROPS IN SLOVAKIA ?

This study investigates the productivity change in the production of cereal and oilseed crops in Slovakia with special emphasis on technical change analysis. It employs a non-parametric distance function approach to measure Malmquist productivity index which is decomposed into technical efficiency change and technical change. Technical change is further decomposed into technical change magnitude and inputand output-bias indices. The productivity change components provide more detailed information about character of productivity change itself and its sources. Our results indicate that productivity in the analysed sector decreased approximately by 20% within the examined period of 1998-2007. The decrease was caused mostly by worsening the technical change (-41.6%). Indices of inputand output bias of technical change were various from unity what suggests that technical change was not Hicks’neutral. Results of further analysis of the direction of technical change bias indicate that farms in average tend to apply fertilizers-using/seed-saving, seed-using/labour-saving, and fertilizers-using/labour-saving technical change bias over the whole sample period, as well as in the EU pre-accession and EU post-accession periods.


INTRODUCTION
In recent years several papers were published on productivity growth in agriculture employing nonparametric methods based on Data Envelopment Analysis (DEA).There are several analyses done on macro data of EU countries (Galanopoulos, Kragiannis, Koutroumanidis, 2004), developed and developing countries (Trueblood, Coggins, 2003;Fulginiti, Perrin, 1997) (Lovell, 2003) is the methodological approach focusing on decomposition of technical change to the components that enable evaluate technical change bias, i.e. proportionality of changes of output isoquants at different mixes of inputs, or proportionality of changes of input isoquants at different output mixes.This property is a prerequisite for assessing whether technical change is Hicks'-neutral, or Hicks'biased.Applications can be found in transportation sector (Barros, Weber, 2009), bank sector (Barros, Managi, Matousek, 2009), in education (Barros, Guironnet, Peypoch, 2011), and in international comparisons (Chen, Yu, 2012).
In this paper an attempt is made to examine the productivity change and its components in the sector of production of cereal and oilseed crops in Slovakia in the period 1998-2007.Special motivation is to learn what is behind the high technical regress.Both, cereals and oilseed crops are cultivated using very similar technology and it is the reason we treat them together.Within the period examined they created ca 75% share on arable land in Slovakia and are considered as a stable part in farm production structure.In the period 1998-2007 Slovak agriculture mostly finished transformation to a market oriented economy and Slovak Republic has joined European Union.Both facts had a significant impact on the farming sector, resulting in the reduction of subsidies from the government budget, more tough international competition, higher food imports, and an access to support funds of EU within Common Agricultural Policy (Bartošová, Bartová, Fidrmuc, 2007).We are trying to link estimated productivity change indicators to the mentioned factors.
The paper itself is divided into five main sections.The second section focuses on the theoretical background to the indexes of productivity and technical change employed.The third section deals with the specification of inputs and outputs employed in the evaluation of technical efficiency and technical change in the sector of cereals and oilseeds.The fourth section presents the resultant indices of productivity, efficiency, and technical change and their components.The paper ends with some brief concluding remarks in the final section.

Malmquist index of total factor productivity change and its components
Malmquist index of productivity change is an indicator enabling to measure productivity change of several factors between two adjacent periods.Malmquist index employs Shephard's distance functions.Output oriented distance function for the period t defined by Shephard (1970) is: where inf is an operator for infimum, θ is a scalar, , … ∈ is a vector of inputs and , … ∈ is a vector of outputs in time period t.
Färe, Grosskopf, Lindgren and Roos (1989,1994) inspired by Caves, Christensen and Diewert (1982) defined output oriented Malmquist index as the geometric mean of two Malmquist indexes for two adjacent periods t and t+1, using reference technology S t , as well as technology S t+1 ., , , With regard to character of employed distance functions Malmquist index , , , ⋛ 1, according as productivity change between two periods t and t+1 can be positive, zero or negative.
According Färe, Grosskopf, Lindgren and Roos (1989, 1994) Malmquist index (3) can be decomposed to technical efficiency change (TECH) and technical (technological) change (TCH).Following Fare et al. (1989Fare et al. ( , 1994) ) an equivalent way of writing this index is: , .TCH , , , where TECH>1 indicates improvement in technical efficiency and TECH<1 deterioration in technical efficiency.TCH>1 indicates technical progress (evidence of innovation) and TCH<1 technical regress.Both components equal unity are associated with no change.
Likewise Malmquist index of total factor productivity change equal unity means stagnation, index greater that unity indicates growth and index less that unity means deterioration of productivity.Malmquist index in (3) a ( 4) is based on the assumption that technology exhibits constant returns to scale (CRS).If the assumption on returns to scale is relaxed to allow variable returns to scale (VRS), then component of TECH in (4), following Färe, Grosskopf, Lovell (1994), can be further decomposed to scale efficiency change (SECH) and pure efficiency change (PECH): Changes in inputs structure in favour of technologically more advanced and effective inputs may lead to biases, which may result in non-proportional shifts of input isoquants.(1989,1994) to suggestion to employ Data Envelopment Analysis to its estimation.
Estimation of distance function values for components calculation needs to apply 8 DEA models for each decision making unit, list of which is presented in Table 2.

Data
Data for the study are drawn from nationally representative sample of the Ministry of Agriculture (information sheets on farms).In the analysis panel data representing 422 farms for the period 1998-2007 has been used in following structure: 104 commercial farms and 338 cooperative farms.
For the purposes of subsequent analysis, we categorise farm data into two groups: data representing EU pre-accession years 1998-2003 and data representing EU post-accession years 2004-2007.Total acreage of the farms examined in the study makes more than 51% of the total arable land in Slovakia.Two output-and three input variables have been used in the estimation of production frontier: -output 1: cereals and oilseed production (tons) -output 2: crop sales (thous.SKK) -input 1: fertilizers costs (thous.SKK) -input 2: seed costs (thous.SKK) -input 3: labour costs (thous.SKK) Table 3 shows descriptive statistics of the variables for the year 2007.Descriptive statistics for all sample years 1998-2006 is presented in the Appendix 1.

RESULTS AND DISCUSSION
In this section, we present summary description of average and cumulative performance indices for all 422 farms within 10 year horizon.Table 4 presents geometric mean estimates of productivity change and its components for the pooled farms by year, geometric mean for the whole period and cumulative indices., 2000, 2003).On the other hand 56.4% productivity increase in 2004 was a result of favorable growing conditions.Only 29% farms of the sample period were able to increase their productivity.
Decrease of productivity was mitigated by improvement of technical efficiency (TECH) by almost 37% within the whole period.Improvement of technical efficiency was probably invoked by more tough competition within the sector and at the market.Almost 82% farms improved their TECH within the sample period.
Decomposition of TECH indicates that improvement of technical efficiency in our sample was caused predominantly by improvement of pure technical efficiency (PECH) -approximately by 27%, and to certain extends by improvement of scale efficiency (SECH) -more that 7%.Improvement in PECH is usually interpreted as an improvement in management of production.SECH increase may be a result of the fact that farms are approaching the optimal scale size for the sector of cereals and oilseed crops production.
Productivity decrease was caused mostly by negative technical change (TCH) -almost 42% within the whole period.It may indicate lack of innovation in production technology, mainly as far as the absence of introduction of new crop varieties resistant to weather extremes, and production processes minimizing impact of negative natural conditions.
The technical change part of the Malmquist index consists of the indices of magnitude (neutral) change, input-biased change and output-biased change of the technology.These components reflect intertemporal movements of the best practice frontier.Our results in Table 4 show that average input-biased technical change equals 1.022.Since it is different form one it indicates that technical change in this sector cannot be assumed Hicks' -neutral.
Cumulative index of magnitude of technical change (MTCH = 0.392) for the sample period indicates significant neutral technology regress.
In the Table 4 we present also comparison of the EU pre-accession period to the EU post-accession period average cumulative indices of productivity and its components.Better results in favour of post-accession period were found as far as the M o , TECH, PECH, TCH, and MTCH.Worse results were found in SECH.All differences are statistically significant.It can lead to conclusion that EU accession had a positive impact on farm performance in the sector.Only scale efficiency change has deteriorated by more than 2 percentage points.
Further we provide results of a more detailed investigation of the direction of technical change through the analysis of the bias direction and input ratios.In Table 5 we summarise the number of farms that experience a bias in the use of inputs.Farms are distributed according to three classes of IBTCH values.Except year 2001/2000 in all years farms with IBTCH >1 prevail.
Recall that in the analysis ratios of three inputs are considered, fertilizers (F), seed (S), and labour (L).There are three combinations of the inputs, F vs. S, S vs. L, and F vs. L to identify the bias direction.With respect to rules in Table 1 if x r /x s increases, then IBTCH>1 implies x r -using bias and IBTCH<1 implies x s -using bias.If x r /x s decreases, then IBTCH>1 implies x s -using bias and IBTCH<1 implies x r -using bias.
According to Table 5 technical change bias indicates that producers generally do not tend to follow any factor using/saving pattern over the examined period.Distribution of farms within the sample period shows significant changes in some years.Average numbers show that majority of farms follow fertilizers-using/seedsaving, seed-using/labour-saving, and fertilizersusing/labour-saving technical change bias.
According to average values of IBTCH and input mix ratios, shown in Table 6, in the sample period farms experience fertilizers-using bias as compared with the use of seed.The same pattern is seen also in preaccession period, as well as in post-accession period.For the input pair of seed versus labour, for all three periods, pattern of seed-using/labour-saving bias is estimated.The last input mix pair -fertilizers vs. labour exhibits fertilizers-using and labour saving bias for all three periods.
Great variability of IBTCH and input bias orientation in year-to-year development does not allow concluding on any statistically significant using-saving pattern.

CONCLUSIONS
The objective of the paper was to analyse productivity change in the sector of cereals and oilseed production in Slovakia and to examine its development from the aspects of its components in the period 1998-2007.that there is great variability of IBTCH and input bias orientation analysis in year-to-year development does not allow concluding on any using-saving pattern.In average for the whole sample period as well as for the pre-accession period, and post-accession period farms tend to apply fertilizersusing/seed-saving, seed-using/labour-saving, and fertilizers-using/labour-saving technical change bias.
, or country provinces (Nin, Arndt, Preckel, 2003).Several works are based on micro data: Latruffe et al. (2012) and Sipiläinen & Kumbhakar (2010) examined productivity change of dairy farms in EU countries; Sipiläinen & Rihänen (2005) focused on silage producers in Finland; Latruffe & Fogarasi (2009) investigated productivity change differences of mixed farms in France and Hungary.In the above referenced studies Malmquist index of total factor productivity change is used as a basic indicator.It is frequently decomposed into technical efficiency change index and technical change index.Both components illustrate what is the source of productivity change, whether it is efficiency catch-up or technological progress as a result of innovation.Widely used is also decomposition of technical efficiency change to pure efficiency change and scale efficiency change, which give an indication of whether farms improve their productivity by better management, or by a shift to the most productive scale size (see e.g.Wu et al, 2001, Lissitsa -Rungsuriyawiboon, 2006).Relatively new and in literature still debated One possible way how to evaluate those changes is to decompose technical change to output bias of technical change (OBTCH) index, input bias of technical change (IBTCH) index and the magnitude of technical change (MTCH) (Färe, Grifel-Tatjé, Grosskopf, Lovell, 1997): TCH , , , Values of Malmquist index, technical efficiency change, pure efficiency change, scale efficiency change, and technical change greater than one indicate productivity gains, increases in efficiency, or technological progress.Values of input-biased technical change, output-biased technical change, and magnitude of technical change different from one indicate that technical change is not Hicks' neutral.
It employs a non-parametric distance function approach to measure Malmquist productivity index which is decomposed into technical efficiency change, scale efficiency change, and technical change.Technical change is further decomposed into technical change magnitude and input-and output-bias indices of technical change.Productivity change components provide more detailed information about character of productivity change itself and its sources.Our results indicate that productivity in the analysed sector decreased approximately by 20% within the examined period.Decrease in productivity was mitigated by technical efficiency improvement, what may indicate positive impact of competition.This improvement was driven mainly by pure efficiency improvement what could be understood as an economy of scale effect.The productivity decrease was caused mostly by worsening the technical change (-42%), what may indicate deterioration of technology and lack of investment into the new technology.Components of technical change -indices of input-and output-bias of technical change were various from unity what suggests that technical change was Hicks' non-neutral.Detailed analysis of input bias of technical change shows

Table 1 :
Input biased technical change and changes in the input mix

Table 2 :
DEA models for distance functions estimation

Table 3 :
Descriptive statistics of the data, year 2007

Table 5 :
Distribution of farms according to year-to-year input biased technical change

Table 6 :
Geometric means of input mix ratios and bias directions Years 1998-2003 represent EU pre-accession period, years 2004-2007 represent EU post-accession period.Source: author's calculations