Reforms and agricultural productivity in Central and Eastern Europe and the Former Soviet Republics: 1989–2005

Reforms and agricultural productivity in Central and EasternEurope and the Former Soviet Republics: 1989–2005

Johan F. M. Swinnen • Liesbet Vranken

Published online: 27 November 2009

� Springer Science+Business Media, LLC 2009

Abstract This paper analyses the changes in agricultural

performance in Central and Eastern Europe and the Former

Soviet republics since the start of the transition process.

We provide a conceptual framework for the evolution of

productivity and efficiency measures and link this evolu-

tion to the issue of factor abundance taking into account

specific transition characteristics. We document the chan-

ges in agricultural performance using empirical data on the

evolution of partial productivity and total factor produc-

tivity estimates and we illustrate how productivity varies

between countries at various stages of the transition pro-

cess. Over the past twenty years, virtually all transition

countries witnessed an initial decline in productivity, and

virtually all countries currently witness an increase in

productivity. However, the depth and length of the initial

decline differs enormously between countries. Our analysis

indicates that the productivity changes were related to the

extent of the pre-reform distortions, initial resource

endowments and technology use, and the reform imple-

mentation in the countries.

Keywords Transition agriculture � Labor productivity �Yields � Technical efficiency � Total factor productivity �Resource endowment � Reform implementation

JEL Classification Q12 � P27 � P32 � R11

1 Introduction

Economic and institutional reforms have dramatically

affected agricultural organization, output, and production

efficiency in Central and Eastern Europe and the Former

Soviet republics. Not only did farm output fall dramatically

in the transition countries of Europe and the former Soviet

Union (FSU), some studies find that efficiency decreased as

well during transition. In a review of the evidence, Rozelle

and Swinnen (2004) conclude that productivity started

increasing early on during transition in Central Europe and

parts of the Balkans and the Baltic states, but continued to

decline much longer in other parts of the FSU. Initial

declines in productivity were associated with initial dis-

ruptions due to land reforms and farm restructuring in

Eastern Europe (Macours and Swinnen 2000a) or with poor

incentives and soft budget constraints in some of the

countries of the former Soviet Union (Sedik et al. 1999;

Lerman et al. 2004) and with disorganization in the supply

chains (Gow and Swinnen 1998).

However, there are several problems in comparing

efficiency studies and drawing implications from them.

First, a limitation is that those studies which include more

countries and a longer time horizon use aggregate data

J. F. M. Swinnen � L. Vranken (&)

LICOS Centre for Institutions and Economic Performance,

University of Leuven (KULeuven), Deberiotstraat 34,

Leuven 3000, Belgium

e-mail: [email protected]

J. F. M. Swinnen

e-mail: [email protected]

J. F. M. Swinnen

Department of Economics, University of Leuven (KULeuven),

Deberiotstraat 34, Leuven 3000, Belgium

L. Vranken

Centre for Corporate Sustainability, HUBrussel, Stormstraat 2,

Brussel 1000, Belgium

L. Vranken

Unit Transition Energy Environment, VITO, Boeretang 200,

Mol 2400, Belgium

123

J Prod Anal (2010) 33:241–258

DOI 10.1007/s11123-009-0162-6

(Mathijs and Swinnen 2001) while studies using farm-level

data are restricted to one country and short time periods,

often even 1 year. Second, cross-country comparisons and

conclusions are complicated by differences in data sam-

ples. Third, linking efficiency changes to specific reforms

is difficult, but important in understanding which factors

have been crucial in constraining or stimulating efficiency

growth.

To get a comprehensive picture of productivity

developments and to accommodate important data con-

straints, we analyze four sets of productivity indicators:

labor productivity (output per unit of labor use) as this

relates to wages; yields (output per unit of land) as this

relates to land rents; aggregate total factor productivity

(TFP) as this relates to the rate of cost-reduction; and

micro-data based total factor productivity (micro-TFP)

based on data envelopment analysis (DEA) as this relates

to potential output that can be generated with the current

input use.

The most comprehensive indicator of productivity is

micro-data based TFP. However because of the absence of

consistent panel data both across countries and over time,

comparative and reliable estimates of micro-TFP analyses

across countries and over time are not consistently avail-

able. For those countries where (even aggregate) TFP

series are available, comparisons have to be done carefully

because of differences in methodologies, time frames,

sampling and commodity coverage. Information for the

partial productivity measures is more readily available, and

so we start by examining indicators of partial productivity

and complement the analysis with TFP estimates.

This paper has seven parts. After this introduction, we

provide in the second part a conceptual framework for the

evolution of productivity and efficiency measures and link

this evolution to the issue of factor abundance taking into

account specific transition characteristics. The third part

summarizes and updates earlier studies and empirical data

on the evolution of partial productivity estimates, such as

labor productivity and yields. The fourth part presents a set

of new total factor productivity estimates. The fifth section

uses farm-level survey data to estimate how productivity

varies between countries at various stages of the transition

process. The sixth section discusses the role of technology

and resource endowments and the nature of the produc-

tivity gains. The final section draws conclusions from the

combination of these pieces of empirical evidence.

For much of the discussion in this paper we will refer to

and compare productivity (changes) in four periods: the

pre-reform period (before 1990), early transition (year 1–5;

roughly the first half of the 1990s), mid transition (year

6–10; the second half of the 1990s), the recent period (after

1999). In Central Europe and the Balkans, the start of the

reforms is assumed to be the year 1989, while in the FSU

(Baltics, European CIS, Transcaucasus and Central Asia)1

the start is assumed in 1990.

2 Resource endowments, reforms, technical change and

productivity in transition: a conceptual framework

There were (and still are) major differences in resource

endowments and the nature of technology in agriculture

among the transition countries. For example, the pre-

reform labor/land ratio in Russia and Kazakhstan was

many times lower than in, for example, Albania, Azer-

baijan, Moldova or Romania (see Table 7). It is well

known that resource endowments (factor proportions) can

play an important role in agricultural productivity growth.

Most famously, Hyami and Ruttan (1970) showed how

technological adaptation can occur through a sequence of

induced innovations in technology biased toward saving

the limiting factors.2 Hence, in countries where land is

relatively scarce and labor abundant, innovations in tech-

nology will be biased towards using land, the scarce factor,

more efficiently—such as through biological innovations

(e.g. seed improvements). In contrast, in countries where

labor is relatively scarce and land abundant, technological

innovations will be induced that contribute to a more

efficient use of labor, the limiting factor. This can come

through, for example, mechanical innovations, which

substitute labor by land and/or capital, leading to a

decrease in the labor/land ratio.

It is also well-known that resource endowments (or

initial technology) also played an important role in the

institutional reforms and productivity changes in transition

agriculture (Macours and Swinnen 2002). However, the

very nature of the transition process, and its inherent policy

and institutional reforms which interact with pre-reform

distortions, makes the impact of resource endowments on

productivity growth considerably more complex than the

original Hayami-Ruttan model (or its more recent exten-

sions) would predict in transition economies.

First, under the Communist system, government regu-

lations caused distortions in many output and factor prices

and in company and household allocation decisions.

Hence, the removal of these distortions in the transition

process caused important readjustments in factor alloca-

tions which primarily reflected pre-reform distortions

1 We refer both to the former Soviet Union (FSU) and to the

Commonwealth of Independent States (CIS), which include all FSU

states except for the Baltic states. Central and Eastern Europe usually

refers to the combination of the Baltic states, the Central European

countries and the Balkan countries (see also Table 1 for a country

classification).2 See e.g. Koppel (1995) for a critical assessment and more recent

interpretations of the Hayami-Ruttan induced innovation model.

242 J Prod Anal (2010) 33:241–258

123

(Liefert et al. 2003). For example, under the central plan-

ning system, labor was inefficiently used in most sectors of

the economy, but this was especially the case for the

agricultural sector (Brada 1989; Bofinger 1993; Jackman

1994). Hence, reforms which allow a more efficient labor

allocation would induce an outflow of agricultural labor

from agriculture to other sectors, and hence a decline in the

labor/land ratio, and the size of this would strongly be

affected by the pre-reform distortions, rather than by

induced technical change.

Second, actual factor adjustments depend on several

factors. One of them is the actual implementation of the

reform policies. In many transition countries, reforms were

implemented with delays, or not, which affected the real-

location of production factors to more efficient uses.

Hence, many of the predicted factor adjustments were

conditional on sufficient progress in the reform process.

Similarly, in order to induce technical innovations to use

land and labor more productively, access to credit for

investment in agricultural machinery and working capital to

buy for example fertilizer and chemicals is crucial. How-

ever, in the early transition years, external credit was scarce

and expensive which resulted in problems accessing phys-

ical inputs (Arnade and Gopinath 1998; Petrick 2004).

Consequently, technical innovations that one would expect

based on the initial resource endowments might not be

observed. Again, progress in reforms was essential to make

this occur, either in making domestic finance available to

farmers and the food industry or by attracting international

capital through foreign direct investment (FDI) in the food

industry and agribusiness with positive vertical spill-overs

on farms. The latter process, which was conditional on the

progress of reforms, has been a very important source of

access to inputs, technology, credit and markets for transi-

tion country farmers, resulting in substantial productivity

growth (Gow and Swinnen 1998; Dries and Swinnen 2004).

Third, another element is that restructuring is a timely

and costly process and factor reallocation will be deter-

mined by opportunity costs in the transition process itself,

which may be affected by government policies and the

level of development. For example, mobility out of agri-

culture is determined by the probability of finding

employment in other sectors and by social benefits that an

individual would receive when leaving agriculture. In

richer transition countries, labor that was laid off (also

from large farms) received sufficient unemployment pay-

ments or pensions to become effectively unemployed or to

retire. In poorer countries, such social payments were

insufficient as an income source for households and the

agricultural sector played a buffer role (‘‘labor sink’’)

during transition, absorbing and providing an income

generating activity to those laid off in other sectors (Seeth

et al. 1998).

Fourth, an important element of ‘‘induced innova-

tion’’—more in line with the broader Hyami and Ruttan

(1985) framework—which strongly affected productivity

change during transition was the endogenous adjustments

of the structure of farms. The relative efficiency of farm

organizations and thus incentives for farm restructuring are

significantly affected by initial resource endowments

because they affect the costs and benefits of shifting to

family farms (from corporate farms). If labor/land ratios

are high, i.e. if agricultural production processes are rela-

tively labor intensive, the benefits of shifting to family

farms (from corporate farms) are larger, while the costs of

shifting are lower (Swinnen and Rozelle 2006). As a result,

particularly land productivity gains which stem from the

shift to household farming are observed. On the other hand,

if labor/land ratios are low, i.e. if agricultural production is

relatively land intensive, the benefits of shifting to family

farms is lower so that large-scale corporate farming

remained. In that case, particularly labor productivity gains

which stem from laying off corporate farm workers will be

observed.

Again, however, it should be emphasized that these

adjustments and these induced farm restructuring processes

were conditional on land reforms and farm privatization

policies being implemented. In many countries, these effects

were delayed (or did not occur) because such reforms were

implemented with delay or not implemented at all.

Finally, an important element of ‘‘induced innovation’’

during transition was the endogenous choice of land

reforms. This effect is beyond the focus of this paper and

we refer to Rozelle and Swinnen (2004) for a discussion

and analysis of this effect.

In summary, resource endowments played an important

role in affecting productivity changes in the transition

process. For example, they affected the incentives for

endogenous farm restructuring and adjustments in factor

proportions. However, these adjustments and the mecha-

nism of productivity growth and technical change were

also strongly affected by the level of development of a

country and its progress and implementation of various

institutional and policy reforms. Moreover, the choice and

implementation of the reforms were often affected by the

level of development and the resource endowments, mak-

ing the set of interactions in determining the pace and

mechanism of productivity growth complex.

3 Partial productivity indicators

3.1 Labor productivity

Agricultural labor productivity (ALP) is measured as out-

put per farm worker. Changes in ALP since the start of

J Prod Anal (2010) 33:241–258 243

123

transition are summarized in Fig. 1 and Table 1. As most

productivity indicators, ALP evolutions differ strongly

among transition countries.

3.1.1 Early and mid-transition

Despite strong decreases in aggregate output (see Table 2),

output per worker rises strongly over the first decade of

transition in Central European countries such as Hungary,

the Czech Republic, and Slovakia. The dramatic reduction

in the use of agricultural labor drives the rise of ALP in

Central Europe (see Table 4). Official employment data

from Central Europe show an average reduction of labor

use of 35% during the first 5 years of transition. The

strongest reductions occur in Hungary (57%) and the Czech

Republic (46%). The same process occurs in Estonia, an

early and radically reforming country, where labor use

-60

-40

-20

0

20

40

60

80

100

120

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Years after start reforms

ALP

Inde

x

Central Europe

Balkan

Baltics

European CIS

Transcaucasus

Fig. 1 Changes in agricultural labor productivity (output per farm

worker—ALP) index in transition countries. Note European CIS

includes Russia and Ukraine and excludes Moldova and Belarus.

Transcaucasus includes Armenia and Azerbaijan and excludes

Georgia

Table 1 Growth of agricultural labor productivity (output per farm worker—ALP) in ECA countries (index = 100 in first year of reform)

Year with

lowest ALP

ALP index in year

of lowest ALP

ALP index

after 5 years

ALP index after

10 years

ALP index after

13/14/15 years

Average annual change

Years

0–5

Years

5–10

Years

10–13/14/15

Central Europe

Czech Republic 1 97 140 198 240 8.0 11.7 8.3

Hungary 1 99 164 207 306 12.8 8.5 19.9

Poland 5 86 86 102 117 -2.9 3.3 2.9

Slovakia 1 100 110 145 233 1.9 7.1 17.5

Balkans

Albania 2 77 108 119 195 1.6 2.1 15.3

Bulgaria 14 57 69 64 65 -6.3 -0.8 0.1

Romania 3 67 79 81 158 -4.2 0.4 15.4

Slovenia 3 61 85 85 96 -3.0 -0.1 2.2

Baltics

Estonia 1 71 138 153 196a 7.7 2.9 10.8a

Latvia 8 54 64 68 75 -7.3 0.9 1.6

Lithuania 4 59 61 73 100 -7.8 2.4 6.7

European CIS

Russia 10 55 63 55 84a -7.3 -1.7 7.2a

Ukraine 9 51 65 55 63a -7.1 -2.0 2.1a

Transcaucasus

Armenia 11 36 42 40 52a -11.5 -0.4 3.1a

Azerbaijan 9 48 57 53 61a -8.7 -0.8 2.2a

Central Asia

Kazakhstan 6 58 60 192 155a -8.0 26.5 -12.5a

Kyrgyzstan 5 58 58 67 74b -8.3 1.7 2.4b

Tajikistan 9 36 46 39 42b -10.7 -1.4 1.0b

Turkmenistan 6 55 88 71 103b -2.4 -3.4 10.6b

Uzbekistan 6 80 88 98 113b -2.4 2.0 5.2b

Other countries have data for 15 years after start of reform. Source: FAO, Asian Development Bank, ILO, OECDa Countries with Estonia, European CIS, Transcaucasus and Kazakhstan have data for 14 years after start of reformb Countries with Kyrgyzstan, Tajikistan, Turkmenistan, Uzbekistan have data for 13 years after start of reform

244 J Prod Anal (2010) 33:241–258

123

declines by 58% within the first 5 years of reform, also

causing an increase in ALP.

In other East European countries, such as Poland, Lat-

via, and Lithuania, ALP falls immediately after reform, but

recovers and rises after the first 4 years. Since then labor

productivity growth has been consistently positive.

However, further East and South, labor productivity

generally falls for much of the first decade of transition.

For example in Russia, Ukraine and Central Asia ALP falls

between 35 and 50% between 1990 and 1999. The most

dramatic fall is in the early years of transition when aver-

age annual ALP declines are more than 7% for several

countries. The productivity decline slows down in the

second half of the 1990s, but the decline continues in many

countries.

In contrast to Central Europe, agricultural labor use

increases in parts of the Transcaucasus and Central Asia.

For example, in Kyrgyzstan agricultural employment

surged between 1990 and 2000, rising by 64% (ILO 2001).

There, as in other countries in the region, a rising number

of people in agriculture coupled with stagnant output led to

the fall in ALP.

The story is mixed in the Balkans. There is a strong ALP

decline in Bulgaria and Romania, and stagnation in the

second half of the 1990s. Albania which has seen contin-

uous ALP growth from the start of transition has more

positive development.

3.1.2 Recent period

Since 1999 there is labor productivity growth in all tran-

sition countries, albeit at quite different growth rates. In

Central Europe and some of the Baltic countries ALP

growth continues from the previous period, with further

outflow of labor from agriculture. In some countries, such

as Hungary, Slovakia, Czech Republic, Estonia and Lith-

uania, productivity growth is very strong (between 7 and

20% average per year). Notice that all these countries have

a significant part of their agriculture organized in large

scale farming companies.

Table 2 Growth of gross

agricultural output (GAO) in

ECA countries (index = 100 in

first year of reform)

Source: FAO

Years after start

of reform with

lowest GOA

GOA index

in year of

lowest GOA

GOA index

after 5 years

of reform

GOA index

after 10 years

of reform

GOA index

after 15 years

of reform

Central Europe

Czech Republic 13 63 75 77 69

Hungary 6 69 70 73 79

Poland 5 77 77 85 90

Slovakia 11 61 77 68 76

Balkans

Albania 2 77 100 113 132

Bulgaria 14 55 63 62 63

Romania 3 75 93 93 117

Slovenia 3 65 81 79 85

Baltics

Estonia 8 41 55 42 45

Latvia 9 37 50 38 48

Lithuania 11 61 69 65 72

European CIS

Belarus 9 57 61 58 69

Russia 8 58 64 62 68

Ukraine 13 50 69 55 58

Transcaucasus

Armenia 3 72 82 80 92

Azerbaijan 5 55 55 72 91

Georgia 10 51 62 51 61

Central Asia

Kazakhstan 8 41 53 52 54

Kyrgyzstan 5 79 79 110 106

Tajikistan 9 48 61 53 71

Turkmenistan 6 69 106 99 160

Uzbekistan 6 90 98 97 119

J Prod Anal (2010) 33:241–258 245

123

In countries dominated by individual farms, such as

Poland, Latvia, and Slovenia, labor productivity growth is

much smaller—reflecting very different labor governance

models on both types of farms (Dries and Swinnen 2002,

2004; Swinnen et al. 2005).

In the rest of the transition region, labor productivity

turned around in the recent period, growing very strongly

in Russia (7%) and Turkmenistan (10%), while growing

more slowly (3% or less per year) in countries such as

Ukraine, Armenia, Azerbaijan, Kyrgyzstan, Tajikistan.3

3.2 Yields and land productivity

The changes in yields are reported in Fig. 2 and Table 3 for

three 5 year periods. The numbers in the tables are sum-

maries of yield evolutions for selected commodities:

grains, sugar beet/ cotton,4 and milk.

3.2.1 Early transition (0–5 years)

Average yields fall during the first few years after reform

for all transition countries (with the exception of Romania

where they are stable). The fall is strongest in the FSU

countries where average annual declines vary from 3.4% in

the Baltics to more than 5% in the rest of the FSU. In

Russia and Ukraine and in Central Asia, yields fall rapidly

during the first 5 years of transition to levels about 25%

lower than pre-reform yields.

3.2.2 Mid-transition (5–10 years)

As in the case of labor productivity, after the initial post-

transition years, the paths of yield evolution differ strongly

between the regions. Central European yields not only

decline less than those in countries further East—by

10–15% on average during the first few years of transi-

tion—they also begin recovering faster (generally from the

third year of transition onward). Agricultural yields

increase, on average, by 3.6% annually in the second half

of the 1990s. A similar, but more pronounced, yield pattern

can be observed in the Baltics. Average yields in the

Baltics dropped initially to almost 25% below their pre-

reform levels. In the second half of the mid 1990s, how-

ever, they recover, rising by an average of 4.6% annually.

In other regions, the recovery is slower and starts later.

In the Balkan countries, the decline in yields is less than in

the Baltic states and more comparable to the decline in

Central Europe: the average total decline is 10–15%. But

their yield recovery is slower, an average of 1.0% annually

during the second half of the 1990s. In the European CIS,

the average growth rate is zero, with positive yield growth

in Russia being offset by declining yields in Ukraine and

Moldova. In Central Asia, growth is strongly positive at

3.2% average, with only Turkmenistan still having

(strongly) declining yields.

3.2.3 Recent period (10–15 years)

Since 1999 yields are improving in all transition regions,

albeit at quite different growth rates. Yields continue to

increase at the same rate (3.5% average annually) in

Central Europe. In the Balkans, yield growth is stronger at

4.5% on average, while somewhat less in the Baltics (3.0%,

down from 4.6% in the previous period). Yield growth

however is strongest in the rest of the FSU. In the European

part of the CIS, yield growth is strong at 5.7%, but is even

outperformed by Central Asia at 7.1% on average.

3.3 Summary

In general, partial productivity indicators follow a J (or U)

curve: initially a decline in productivity is observed fol-

lowed by a recovery later on. However, while productivity

generally fell following the reforms which were launched

in the late 1980s and early 1990s, the extent of the decline

(i.e. its depth and length) and the strength of the recovery

differs strongly between countries.

The productivity indicators presented above suggest that

the productivity changes were related to the extent of the

pre-reform distortions and the reform implementation in

the countries. In the most advanced reformers (mostly in

Central Europe) the decline was relatively mild and

-35

-30

-25

-20

-15

-10

-5

0

5

10

15

20

25

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Ave

rag

e ag

r. y

ield

ind

ex

Central Europe

Balkans

Baltics

European CIS

Fig. 2 Change in average agricultural yields. Agricultural yield index

is calculated as the average yield index of grains and milk. Given the

sensitivity of grain yields to the weather conditions, the figures show

the moving average over 3 years. Balkans includes Albania, Bulgaria,

Romania and excludes Slovenia

3 Kazakhstan changed its definitions of agricultural labor making

comparisons across time impossible with the official data.4 Cotton for Central Asia and sugar beets for other countries.

246 J Prod Anal (2010) 33:241–258

123

recovery started fairly early on, and productivity growth

has been strong for about a decade. In the Baltics, recovery

started relatively quickly, reflecting the fast pace of the

reforms, but the initial fall in productivity was much deeper

than in Central Europe, possibly reflecting the larger pre-

reform distortions as these countries were part of the FSU.

In the European CIS countries, the decline of productivity

was much more dramatic than in Central Europe and lasted

for much of the 1990s. It was only after the Russian crisis

that a recovery started, and although hesitant at first, pro-

ductivity increases are strong, certainly for yields.

However, within countries we see quite different evo-

lutions between different partial productivity indicators

(e.g. labor and land productivity). To complement the

partial productivity analysis, we next turn to an analysis of

total factor productivity indicators.

4 Total factor productivity

4.1 Methodology and approach

There are two published cross-country studies which esti-

mate TFP for ECA countries. Macours and Swinnen

(2000b) estimate TFP indices for crop production in CEE

for the period 1989–1995 and Lerman et al. (2004) esti-

mate TFP indices for agriculture as a whole for the CIS

countries. In this paper we extended the study of Macours

and Swinnen (2002) in two ways: we added livestock

output to have total agricultural output and we extended the

series by 6 years, to 2001.

The calculation of total factor productivity is based on

the growth accounting approach as in Macours and Swinnen

(2000a, b). The growth accounting literature empirically

Table 3 Growth of index of agricultural yields in ECA countries (index = 100 in first year of reform)

Total Grainsa Sugarbeet/cotton Milk Average agr.b Average agr. per year

5 10 15 5 10 15 5 10 15 5 10 15 0–5 5–10 10–15

Central Europe

Czech Republic 87.0 89.0 112.4 102.1 128.8 142.2 99.8 126.1 135.7 96.3 114.7 130.1 -0.7 3.7 3.1

Hungary 72.3 82.7 103.4 72.3 100.9 128.9 95.3 110.3 112.9 80.0 98.0 115.1 -4.0 3.6 3.4

Poland 79.7 93.3 110.0 85.7 99.4 125.9 95.7 107.7 128.4 87.1 100.1 121.5 -2.6 2.6 4.3

Slovakia 88.9 88.5 95.3 99.1 118.2 132.1 89.5 115.6 151.9 92.5 107.4 126.4 -1.5 3.0 3.8

Balkan

Albania 85.0 85.8 106.7 72.5 77.3 77.5 125.4 138.2 183.3 94.3 100.4 122.5 -1.1 1.2 4.4

Bulgaria 63.2 64.6 86.1 56.5 68.2 99.0 85.7 90.2 82.2 68.5 74.3 89.1 -6.3 1.2 3.0

Romania 84.6 92.7 104.2 80.0 81.5 121.8 136.9 133.6 183.4 100.5 102.6 136.5 0.1 0.4 6.8

Slovenia Na Na Na 97.2 92.7 98.4 99.5 112.1 128.9 NA NA NA NA NA NA

Baltics

Estonia 69.0 88.1 111.2 102.7 Na Na 86.2 111.9 131.9 86.0 100.0 121.6 -2.8 2.8 4.3

Latvia 71.3 94.0 118.0 88.3 107.4 128.8 89.4 116.5 129.6 83.0 105.9 125.5 -3.4 4.6 3.9

Lithuania 60.7 89.9 95.7 99.6 111.7 133.4 81.0 92.6 81.5 80.5 98.0 103.5 -3.9 3.5 1.1

European CIS

Belarus 74.5 69.0 86.8 66.3 88.6 98.7 76.7 70.4 99.7 72.5 76.0 95.1 -5.5 0.7 3.8

Moldova 82.4 57.5 75.8 NA NA NA 51.4 54.1 81.8 NA NA NA NA NA NA

Russia 62.6 77.3 95.2 79.6 85.2 124.3 74.4 83.6 99.0 72.2 82.0 106.1 -5.6 2.0 4.8

Ukraine 69.7 54.1 72.5 88.0 76.0 106.5 77.0 80.9 140.0 78.2 70.3 106.3 -4.4 -1.6 7.2

Central Asia

Kazakhstanc 41.4 73.8 77.9 79.3 67.0 84.8 NA NA NA 60.3 70.4 81.3 -7.9 2.0 2.2

Kyrgyzstanc 56.5 84.7 95.3 79.0 99.6 119.0 NA NA NA 67.8 92.2 107.1 -6.4 4.9 3.0

Tajikistanc 65.6 107.5 150.4 52.7 52.5 63.4 NA NA NA 59.2 80.0 106.9 -8.2 4.2 5.4

Turkmenistanc 82.4 69.2 122.6 79.3 53.9 50.5 NA NA NA 80.8 61.6 86.6 -3.8 -3.9 5.0

Uzbekistanc 99.5 163.1 227.3 96.2 77.2 96.9 NA NA NA 97.9 120.1 162.1 -0.4 4.5 8.4

Sources: USDA for grains; sugar beet yields are from FAO for Central Europe, Balkans and China, and from ZMP and FAO for Central Asia,

Transcaucasus and European CIS. Milk yields are from ZMP and FAOa Grains include wheat, rice (milled weight) and coarse grainsb Average agricultural yields are calculated as a simple average of the yields of grains, sugar beet/cotton and milkc Central Asia: Cotton instead of sugar beet; Average agr. is average of grains and cotton only

J Prod Anal (2010) 33:241–258 247

123

accounts for growth in output by measuring factor inputs

and an unexplained residual, which is generally attributed to

technological change. More specifically, total factor pro-

ductivity is measured as an output-input ratio where the

estimated coefficients of the different production factors in

a Cobb-Douglas production function are used as input

weights. The output is the net agricultural production index

number (PIN) reported by the FAO. The index equals total

agricultural production excluding the amount of produce

used as animal feed and seed. The index is calculated by the

Laspeyres formula:

Pqtpo

.P

qopo;

where the net production quantity of each commodity

produced in the current year (qt) is weighted by the 1999–

2001 average per unit international commodity prices (po)

and summed for each year. To obtain the index, the

aggregate for a given year is divided by the average for the

base period 1999–2001 where qo is the net production

quantity in the base period (FAO 1986). To ensure com-

parability, a single set of international commodity prices

was used by the FAO for all countries and country groups.

As input factors, we included land, labor, capital, fer-

tiliser and animal stock. As input weights, we used the

coefficients of production function estimated in Cungu and

Swinnen (2003).5 The input variables and data were

obtained as follows:

• Land: following the FAO definition, land includes

arable land and land under permanent crop cultivation,

excluding that left idle.6 The area under permanent

crops includes also land under flowering shrubs, fruit-

trees, nut trees and vines, but excludes land under trees

grown for wood.

• Labor is the number of people who are economically

active in agriculture including those that are either

engaged or seeking employment in agriculture. For the

Central Asian republics, the data comes from the Asian

Development Bank. For the other countries in our

sample, the information has been put together from a

variety of sources including the International Labour

Organisation (ILO), the Organisation for Economic

Cooperation in Europe (OECD), the FAO, and indi-

vidual country reports.7

• Livestock data are also from the FAO database. We

take account of the number of cattle indicated as the

number of live animal heads in a country at the time of

enumeration, except for chicken.

• Capital is based on the physical number of tractors in

use, as reported by the FAO. We do not have

information on the power of those tractors for each

country/year. Therefore, the assumption here is that, in

each country, the type of tractors in terms of their

horsepower stayed the same throughout the entire

period. As such, the horsepower as a conversion factor

drops out when the yearly to 1989 ratios are calculated.

• Fertiliser is the quantity of pure nutrient fertiliser in

metric tonnes consumed in agriculture by a country, as

reported by the FAO. Three types of fertilisers have

been aggregated to produce this measure: nitrogenous,

phosphate, and potash fertilisers.

4.2 Results

We estimate TFP for eight Central and Eastern European

countries: Czech Republic, Hungary, Poland, Slovakia

(Central Europe) and Albania, Bulgaria, Romania and

Slovenia (Balkan). Figure 3 shows the average TFP esti-

mates for the two groups of countries. Tables 2, 4 and 5

provide more country details on the evolution of output,

input use and TFP estimates. The estimated TFP indicators

suggest some remarkable developments.

Over the entire period 1989–2001, TFP growth was

strong in half the countries. Albania, Romania, Hungary

and Slovakia all have average annual growth rates of

around 2.5% (between 2.3 and 7.5%). The others per-

formed much weaker. Czech Republic and Poland have an

average positive TFP growth, but less than 1%. Bulgaria

and Slovenia even have negative average growth.

In terms of dynamics, TFP in Central Europe started

increasing from the start of the reforms, and continued to

increase at rapid rates during the first decade. This was

especially the case in Hungary, Czech Rep and Slovakia

where average TFP growth was positive from the very

80

85

90

95

100

105

110

115

120

125

130

1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002

TF

P

Balkan CEEC

Fig. 3 Total factor productivity change between 1989 and 2002

5 Coefficients of the production function are 0.23 for land, 0.19 for

labor, 0.42 for livestock, 0.09 for capital and 0.23 for livestock.6 Land under permanent crops includes land cultivated with crops

that occupy the land for long periods and need not be replanted after

each harvest.7 See Macours and Swinnen (2000a, b) for full references to these

country sources.

248 J Prod Anal (2010) 33:241–258

123

beginning, and grew strongly at around 3% per year in the

1992–1995 period and around 4% per year in the 1995–

1998 period. In all other countries, including Poland, there

was an initial decline in TFP.

What is remarkable is that in Central Europe, after robust

TFP growth through the 1992–1998 period, TFP growth has

slowed down in these countries, and has even gone negative

in Hungary and the Czech Republic over 1998–2001. To

understand why this is the case, we looked into the partial

productivity indicators. In the early transition period (first

5 years of transition), the use of all inputs, except for land is

decreasing and the decrease is faster than the decrease

in output (Tables 2, 4). In the mid transition period

(5–10 years after the start of the reforms), labor use con-

tinued to decline, while tractor and fertilizer use started to

increase and agricultural output more or less stabilized

(Tables 2, 4). As a result, labor productivity keeps

improving through the period, but output per tractor and

output per unit of fertilizer goes down, reflecting substantial

increases in investments in tractors and more fertilizer use.

In the Balkans the TFP evolutions are also mixed.

Bulgaria has performed worst of all, having recorded

Table 4 Growth of index of input use in ECA countries (index = 100 in first year of reform)

Years after start

of reform

Animal stock Fertiliser Labor Land Tractors

5 10 15 5 10 12/13a 5 10 14/15b 5 10 13/14c 5.0 10.0 13/14c

Central Europe

Czech Rep. 69.0 52.9 45.6 29.4 23.5 32.3 53,7 38,8 31,8 103.3 103.5 103.2 58.1 71.2 82.3

Hungary 58.9 51.4 43.5 14.7 20.3 23.6 42,6 35,1 26,6 94.4 95.4 90.5 71.8 72.0 88.5

Poland 81.2 69.2 56.5 35.4 37.8 37.4 89,3 82,9 77,1 99.4 98.0 85.9 113.7 113.3 119.0

Slovakia 65.1 46.2 35.4 17.0 15.3 21.4 70,6 47,0 32,8 99.8 99.7 99.5 88.8 64.7 61.2

Balkans

Albania 121.3 106.5 96.8 18.6 13.3 44.4 92,3 95,7 67,9 101.4 103.2 101.0 74.4 67.0 70.1

Bulgaria 47.3 42.3 45.9 24.9 13.6 18.4 92,3 95,5 96,9 97.6 90.0 84.4 69.0 74.7 57.7

Romania 62.6 49.7 45.8 27.1 17.1 23.6 118,2 115,1 74,2 100.3 100.1 99.7 106.3 108.0 111.5

Slovenia 86.2 81.8 81.3 56.4 52.2 46.4 95,3 93,4 88,5 90.9 82.8 84.4 56.2 71.7 69.8

Baltics

Estonia 50.4 32.1 30.0 16.9 21.2 16.3 39,7 27,5 22,5 107.4 106.9 89.8 105.8 108.4 115.7

Latvia 37.7 25.9 25.4 21.3 48.6 48.4 78,8 56,4 55,8 99.1 96.8 96.3 82.2 90.9 90.1

Lithuania 52.1 40.6 35.8 10.3 13.3 16.9 112,6 88,8 77,6 100.1 99.5 99.4 118.3 138.1 138.5

European CIS

Belarus 79.5 63.6 58.3 24.6 37.9 35.2 85,6 NA NA 97.6 96.7 92.9 92.1 57.9 46.7

Moldova 64.4 32.2 25.2 42.4 1.9 3.7 113,9 NA NA 102.2 101.8 101.1 92.8 77.7 76.2

Russia 73.6 46.7 38.3 11.2 9.1 9.4 100,2 113,1 80,1 98.3 98.6 98.2 81.5 57.9 45.4

Ukraine 75.2 40.7 26.6 24.3 12.0 16.1 106,1 100,3 91,0 99.6 98.5 98.4 91.7 62.3 76.4

Transcaucasus

Armenia 56.6 53.8 64.5 10.8 10.8 17.4 193,7 199,2 182,9 101.9 108.0 108.0 119.1 117.0 126.8

Azerbaijan 81.7 101.2 122.0 NA NA NA 97,3 137,0 135,5 96.1 103.0 106.1 90.4 90.6 90.8

Georgia 49.3 56.7 63.2 22.3 30.2 20.4 86.2 84.9 85.0 70.6 73.2 93.0

Central Asia

Kazakhstan 77.2 38.2 49.6 15.7 6.1 10.7 88,5 27,0 33,9 96.4 93.0 93.5 78.3 22.8 23.0

Kyrgyzstan 60.6 61.4 68.2 NA NA NA 134,8 164,0 146,0 99.2 101.5 102.0 99.2 101.7 87.7

Tajikistan 86.5 75.2 94.5 29.5 3.8 11.8 131,5 134,5 151,3 97.3 94.1 92.9 84.0 84.0 84.0

Turkmenistan 123.9 89.2 129.0 39.6 31.9 29.5 121,2 140,1 153,5 73.1 73.6 74.4 79.7 54.2 66.4

Uzbekistan 116.1 111.6 114.4 34.1 52.6 51.7 111,7 99,0 98,1 88.9 88.8 87.7 94.4 94.4 94.4

Sources: FAO, Asian Development Bank, International Labor Organisation, OECDa Fertilizer: Data for 12 years after the start of reform for countries in Central Europe and the Balkans and data for 13 years after the start of the

reform for the remaining countriesb Labor: Data for 15 years after the start of reform for countries in Central Europe and the Balkans and data for 14 years after the start of the

reform for the remaining countriesc Land and Tractors: Data for 14 years after the start of reform for countries in Central Europe and the Balkans and data for 13 years after the

start of the reform for the remaining countries

J Prod Anal (2010) 33:241–258 249

123

negative TFP growth for most of the 1989–2001 period,

except for the 1992–1995 period. Romania shows a process

of ups and downs. Periods of substantial TFP declines

(1989–1992 and 1995–1998) interact with periods of very

strong TFP growth (1992–1995 and 1998–2001).

Slovenia and Albania have remarkable TFP records, for

opposite reasons. While Slovenia is a member of the EU

now, and the richest country of all ECA, its average agri-

cultural TFP growth has been negative over the 1989–2001

period, showing negative TFP growth again in the 1998–

2001 period. In stark contrast, Albania, the poorest country

included in this analysis, has, after a small initial decline in

productivity before 1992, strong TFP growth since 1992

when the land reforms and farm restructuring really took off.

5 Farm-level productivity analysis

5.1 Approach and interpretation

To further study the differences in agricultural productivity

we relate them to aggregate changes at the farm level by

analyzing agricultural productivity at the farm level. Our

analysis uses micro data from a set of farm survey data from

eight transition countries (three Central European countries

(Czech Republic, Hungary, Slovakia), two Balkan countries

(Albania and Bulgaria), and three countries from various

parts of the CIS (Moldova, Azerbaijan and Kazakhstan)). We

first calculate farm-level efficiency indicators using data

envelopment analysis (DEA) and calculate kernel density

estimates for each of the countries. We compare efficiency

distributions between countries and calculate aggregated

(country-level) efficiency estimates. We then correlate these

with various indicators of reforms. We discuss how the share

of farmers producing efficiently changes during transition

and which aspects of the reforms are important in explaining

differences in the efficiency distributions of the countries.

It is important to notice that, unless stated explicitly, we

compare differences in efficiency distributions between

countries. Our approach is focused on measuring the rela-

tive efficiency of various farms compared to the most effi-

cient farm within the same country. Based on this we can

estimate an average efficiency level for each country, but

this measure reflects mostly the distribution of farms within

the country rather than a comparison to an external standard

which is the same for all countries. In other words, we do

not compare the efficiency of a specific farm in Kazakhstan

to the efficiency of a specific farm in Hungary—but we do

compare the relative efficiencies of all farms in Kazakhstan

with the relative efficiencies of all farms in Hungary.

This approach means that we do not assume that all

farms across these very different countries have access to

the same technology. We use a much less restrictive

assumption that all farms within a country have access to

the same technology. What is important is that we think

that cross-country comparisons of these distributions are

still relevant because the relative efficiency distributions do

give an important insight into the extent to which the vast

majority of farms have (not) been able to catch up with the

most efficient farms within the country, using the local

technology, know-how, infrastructure.

It is important to realize that, although there will always

be heterogeneity in efficiency among farms in any country,

one would expect that in a country’s transformation pro-

gress to a market economy the gap between the most and

the least efficient farm would decrease, and that the bulk of

the farms would become relatively close to the most effi-

cient farm. As we will show below this is still very dif-

ferent in the various transition countries analysed here.

A critique of our approach could be that in today’s world

most countries have access to the same technology. To

allow for this possibility we also use a more restrictive

assumption and we re-estimated the model assuming a

multi-country frontier and we compare the original

Table 5 Annual growth rates

of total factor productivity for

agriculture in the Balkans and

Central and Eastern Europe for

selected years (%)

Source: own calculations

Average annual change 1989–2001 1989–1992 1992–1995 1995–1998 1998–2001

Albania 2.6 -1.1 5.6 2.1 3.9

Bulgaria -0.4 -1.3 4.0 -4.1 -0.2

Romania 2.5 -4.2 11.6 -4.8 7.5

Slovenia -0.4 -9.9 9.0 1.6 -2.2

Balkans 1.1 -4.1 7.5 -1.3 2.3

Czech Republic 1.4 1.3 2.3 3.9 1.6

Hungary 4.0 1.9 3.4 5.1 5.6

Poland 0.8 -1.7 0.5 3.3 0.9

Slovakia 2.2 0.1 2.4 4.3 2.1

Central Europe 2.1 0.4 2.2 4.2 1.7

All 1.6 -1.9 4.9 1.4 2.0

250 J Prod Anal (2010) 33:241–258

123

efficiency estimations with the efficiency scores relative to

this more general frontier.

5.2 Data

Our analysis with micro data uses a set of farm survey data

from eight ECA countries. The data are a combination of

two different sets of data. The first is of five Central and

East European countries (Albania, Bulgaria, Czech

Republic, Hungary, and Slovakia) and was collected in the

1997–2000 period and based on a common set of survey

instruments—all surveys were coordinated by the same

experts (from the University of Leuven and the University

of Athens). The surveys in Hungary and Bulgaria were

implemented in 1998 and have representative data for

1997. Data for Albania, Czech and Slovak Republics are

for the production year 1999. The second set of surveys

was jointly organized under a 2003–2004 World Bank

project and has survey data from three CIS countries:

Moldova, Azerbaijan and Kazakhstan.

To increase the accuracy of comparisons, we take only

crop farms into consideration to enhance the homogeneity

of the dataset. To be included in our analysis the value of

grain production in the value of total production needed to

be more than 50%. Selection of farms out of the total

sample occurred according to objective criteria and did not

occurred randomly. Further, all country data were checked

for outliers and observations with incomplete information

were dropped.

The cleaning of the data resulted in a dataset of 178

Hungarian farms (63 cooperatives, 40 companies and 75

family farms), 93 Bulgarian farms (45 cooperatives, 9

companies and 39 family farms), 183 Czech farms (38

cooperatives, 14 companies and 131 family farms), 210

Albanian family farms, 783 Azerbaijan farms (15 corporate

farms and 768 family farms), 800 Kazakh farms (22 cor-

porate farms and 778 family farms) and 700 Moldovan

farms (24 corporate farms and 676 family farms).

The Albanian, Hungarian and Bulgarian datasets are

representative for the whole country, while in the Czech

and Slovak Republics some regions were selected for

surveying, but we selected regions with significant varia-

tions in the location of the farms (hills, low land and more

urban areas). In Azerbaijan, Kazakhstan and Moldova, the

datasets are representative within selected regions. There-

fore, the main farming system zones in the country were

first identified and then representative districts were

selected from these main zones. Though the type of agri-

culture practiced was the most important factor in the

identification of these zones, there was also an effort to

ensure coverage of a variety of farm types, different

degrees of poverty and various degrees of isolation (Dud-

wick et al. 2005).

The countries differ largely in terms of income levels,

reforms, land use, and economic conditions (see e.g. the

reform indicators in Table 6). Most countries have a

fragmented land ownership structure due to land restitution

or distribution processes implemented in 1990s. However

in all countries the land reform process was well advanced

by the time of the survey.

In Albania, the poorest country in Europe, almost half of

the active population is still employed in agriculture, and

virtually all agricultural land is cultivated by small indi-

vidual farms. Also in Azerbaijan, land is mostly used by

individual farms. Slovakia and the Czech Republic are the

opposite of Albania and Azerbaijan in most respects. They

are much richer and only around 5% of employment is in

agriculture. The vast majority of the land is used by large-

scale farming companies, successor organizations of for-

mer collective and state farms. The other countries repre-

sent a more mixed structure, albeit with considerable

differences among them. In Hungary, Bulgaria, Moldova

and Kazakhstan land is used by a mixture of large-scale

farming companies and small scale individual farms, with

much regional variation. Hungary has a considerably

higher level of income than Bulgaria and Kazakhstan.

Rural Moldova is the poorest region.

5.3 Methodology

To investigate how average efficiency and the distribution

of efficiency scores have changed during various stages of

transition, we first calculate farm level total technical

efficiency scores using Data Envelopment Analysis

(DEA) for each country. To measure technical efficiency

requires the specification of a frontier production func-

tion, and the measurement of the deviation or distance of

the farms from the frontier, which is then a measure of

technical inefficiency. The technique of DEA constructs a

convex hull around the observed data (Charnes et al.

1978). As in Fare et al. (1985), we assume that production

is characterized by a non-parametric piecewise-linear

technology, so that simple linear programming techniques

can be used to calculate efficiency. We further assume

strong disposability of outputs and inputs and estimate the

non-parametric deterministic frontier, expressed in terms

of minimizing input requirements. Input-oriented techni-

cal efficiency for each farm i (i = 1, 2, …, N) can be

calculated by solving the following linear programming

problem:

minz;k

k

s:t: zY � yi; zX� kxi; z� 0

where k is the efficiency measure, Y is the 1 9 N vector of

observed outputs, yi is the output of farm i, X is the matrix

J Prod Anal (2010) 33:241–258 251

123

of observed inputs, xi is a vector of inputs employed by

farm i and z is a vector of intensities that characterizes each

farm. A farm displays total technical efficiency if it pro-

duces on the boundary of the production possibility set, i.e.

it maximizes output with given inputs and after having

chosen technology. This boundary or frontier is defined as

the best practice observed.

The data used in the DEA calculations are similar for all

countries and include gross output, expressed in local

currency, and data on land, labor, capital and other variable

inputs. Output is the value of physical production valued at

fixed prices. These fixed prices are calculated based on the

price information in the survey. Labor is expressed in

annual working units which correspond to 2,150 labor

hours or the number of hours that a full-time worker can

perform in 1 year. Land is the total amount of agricultural

land cultivated. To take into account quality differences in

land, the area cultivated is multiplied by a land quality

indicator. The value of estimated farm buildings and

agricultural machinery is used as a proxy for capital. Fur-

ther, we also take into account the amount of money spent

on the purchase of seeds, feed, grains, roughage, concen-

trated feed, fertilizers, energy and services. Different

frontiers are assumed for each country.

When calculating efficiency scores using a multi-coun-

try frontier, we expressed output in Euro’s using 2003 as

base year8 and took only two inputs (land and labor) into

account. Quality differences could not be accounted for as

different indicators were applicable across the different

countries. The value of farm buildings and agricultural

machinery as well as the amount spent on variable inputs

could also not be included when calculating the efficiency

Table 6 Efficiency and reform indices of 8 transition countries

Albania-

1999

Bulgaria-

1997

Slovakia-

1999

Czech-

1999

Hungary-

1997

Azerbaijan-

2003

Kazakhstan-

2003

Moldova-

2003

Correlation coefficient

with

ToE single-

country

frontier

ToE multi-

country

frontier

Average ToE, single-

country frontiera25 (35) 37 (24) 41 (25) 43 (25) 47 (27) 30 (26) 21 (20) 31 (24)

Average ToE, multi-

country frontiera10 (12) 13 (12) 23 (18) 23 (16) 16 (11) 9 (10) 7 (10) 3 (3)

EBRD reform indices 2.6 2.9 3.3 3.5 3.7 2.7 3.0 2.8 0.82** 0.72***

Price Liberalisation 3 3 3 3.3 3 4 4 3.7 -0.60 -0.60

Forex and trade

liberalisation

4 4.3 4.3 4.3 4.3 3.7 3.3 4.3 0.78** 0.48

Enterprise reform 2 2.3 3 3.3 3 2 2 2 0.88*** 0.92***

Competition policy 1.7 2.3 3 3 3 2.3 2 1.7 0.86*** 0.89***

Small scale privatisation 4 3.3 4.3 4.3 4.3 3.7 4 3.3 0.38 0.70*

Large-scale privatisation 2 3 4 4 4 2 3 3 0.78** 0.71*

Banking sector reform 2 2.7 3.3 4 2.7 2.3 3 2.3 0.51 0.76**

Reform non-banking

financial institution

1.7 2 3 3.3 2.3 1.7 2.3 2 0.58 0.81**

WB agr reform indices 6.8 5.4 7.6 8.6 8.6 6.6 6 6 0.70* 0.73**

Price and market 8 6 7 9 9 8 6 6 0.52 0.53

Land reform 8 7 8 8 9 8 5 7 0.70* 0.51

Agro-processing 8 5 8 9 9 6 7 6 0.44 0.62

Rural finance 5 4 8 9 8 6 7 7 0.49 0.51

Institutional 5 5 7 8 8 5 5 4 0.81** 0.87***

GDP 96 1,349 4,180 5,206 4,662 866 1,671 371 0.81** 0.87***

Source: EBRD, WB, Own calculations

*, **, *** Indicate whether the correlation coefficient is significantly different from zero at a 10, 5, 1% significance levela Standard deviation between brackets

8 Exchange rates and Consumer Price Indices (CPI) were obtained

from United Nations Economic Commission for Europe (UNECE)

Statistical Division Database which is compiled from national and

international (EUROSTAT and IMF) official sources.

252 J Prod Anal (2010) 33:241–258

123

scores using a multi-country frontier as the list of items

within these categories differed among countries.

5.4 Results of efficiency calculations

The DEA calculations yield efficiency scores for all the

farms which can then be used to calculate an efficiency

distribution for all the countries. Based on the efficiency

distribution for each country, we estimate a kernel density

function. We can calculate the average total technical

efficiency for each country assuming these density func-

tions (see the first line of Table 6).

The DEA calculations show that there are large differ-

ences in efficiency distribution between the countries. In

Kazakhstan, for example most farmers (80%) have an

efficiency score lower than 30 and only a very small share

(2%) of the farmers are achieve an efficiency score close to

100. On the other hand, we observe that in Hungary most

farmers have an efficiency score between 40 and 70 and

9% have an efficiency score between 90 and 100.

The differences in efficiency distribution between

countries are also reflected in the average efficiency scores

(Table 6). The highest average efficiency is in Hungary in

1997 with 47, which means that, on average, farms obtain

47% of the maximum efficiency possible. In Hungary, most

farmers have an efficiency score between 40 and 70 and

9% have an efficiency score between 90 and 100. Also

Slovakia and the Czech Republic do relatively well, with

higher averages than 40. The poorest efficiency indicators

are for Kazakhstan (21), and Albania (25).

The efficiency scores which are calculated using a multi-

country frontier (see the second line of Table 6) are gen-

erally consistent with the previous conclusions. The aver-

age efficiency score of Czech, Slovakian and Hungarian

farmers are higher than those of the Bulgarian farmers and

much higher than those of the Albanian, Azeri, Kazak and

Moldovan farmers—all measured for their respective sur-

vey years.

5.5 Reforms and efficiency

To see whether these estimated differences in country-level

efficiency are related to reforms which have been imple-

mented we compare the efficiency indicators of the coun-

tries with indicators of reforms, from the World Bank and

the European Bank for Reconstruction and Development

(EBRD).9 While one should be careful to draw conclusions

from such cross-comparisons with these data, and in par-

ticular in linking them with country-level indicators, the

correlation estimates in Table 6 and further comparisons

suggest that a country which is further advanced in reform

implementation and in the transition stage has more farms

that can be found on the boundary of the production pos-

sibility set and that the farms reach on average a higher

efficiency level. There is a positive correlation between the

progress of reforms implemented in a country and the

average efficiency level reached by the agricultural pro-

ducers. In countries which are less advanced in the tran-

sition process, there are relatively more inefficient

production units. In countries more advanced in transition,

there are relatively fewer efficient farms.

These observations are in line with our conceptual

arguments in Sect. 2 and the data in the other sections

where productivity growth started first and was strongest in

countries which were most advanced in the implementation

of reforms. Other studies, such as Lerman (2008), also

conclude that agricultural growth and performance are

positively linked with various measures of policy reforms.

While the correlations between the aggregate reform

indicators suggest an important causal affect, the indicators

as such tell us little about the mechanism. Therefore, let us

take a closer look at the correlations between efficiency

scores and the various reform indices. The first observation

is that there is a closer correlation with the EBRD index (a

non-agricultural index) than with the WB agricultural

reform index. This suggests that the key factor may not be

specific to agriculture. One important factor is that all these

surveys were done in countries, and at times, when farms

used private land plots and faced hard budget constraints.

Hence, in these situations, other factors, such as access to

input and output markets are likely to become the prime

determinants of efficiency.

Second, if we disentangle the reform indices and cor-

relate them with the observed efficiency scores (Table 6),

we see that there is significant correlation between effi-

ciency and competition policy, enterprise reform, and

institutional reforms. Again these correlations indicate the

importance of general institutional reforms and reforms of

the sectors ‘‘surrounding agriculture’’ as a source of effi-

ciency growth. In general, good competition policy to

reduce abuse of market power is beneficial for the

9 The World Bank agricultural reform index is an aggregate index of

progress in land reform, price and market liberalization, reforms in

the agro-processing sector and rural finance and of the institutional

reforms. A score of one means no reform, i.e. a situation comparable

with a centrally planned economy. The maximum score that a country

can reach is 10 which means the market reforms have been completed

Footnote 9 continued

and the situation is a free market economy. The EBRD transition

indicator gives a score from 1 to 4. It aggregates assessments of the

privatization of small- and large scale enterprises, enterprise

restructuring, price liberalization, trade and foreign exchange system

liberalization, competition policy, bank and non-bank financial sector

reforms. A rate of 4? is given when standards and performance are

comparable with those of advanced industrial economies. The general

EBRD indicator is the average of the score given to the reforms in

each area.

J Prod Anal (2010) 33:241–258 253

123

performance of an industry. However, in agriculture there

is little market power. Therefore, maybe most important

though is its indirect impact on agricultural producers. It

may have an important impact on firms up- and down-

stream such as agribusiness and food processors. Domi-

nation of large companies in the in- or output markets will

strongly affect farms. Enterprise reforms which contributed

to significant and sustained harder budget constraints and

to promoting corporate governance (e.g. through privati-

sation combined with tight credit and subsidy policies and/

or enforcement of bankruptcy legislation) may also cause

higher efficiency of the farms. Remarkably Table 6 indi-

cates that there is no significant correlation with price and

market reforms, but much more with institutional reform.

Elsewhere (in Swinnen and Vranken 2005), we have

developed a theoretical model to provide some hypotheses

to explain these correlations, and to identify the mecha-

nisms through which general institutional reforms and

factor market imperfections in labor and capital can be the

prime determinants of farm efficiency change during

transition. According to our model, as transition progresses

two developments occur (simultaneously or not) which are

likely to cause an increase in average agricultural pro-

ductivity. The first mechanism is through the labor market.

Labor opportunity costs increase either due to improved

off-farm employment or with improved governments’ fis-

cal situations leading to improved social benefits (unem-

ployment benefits and pensions).10 If returns to labor

outside agriculture relative to the returns to agricultural

labor are higher (lower) for efficient than inefficient pro-

ducers, the most efficient (inefficient) producers might be

inclined to leave agriculture. This results in a narrowing of

the gap between the least and most efficient producers. The

second factor is that other rural factor markets improve,

because of a reduction in transaction costs or improvement

in institutions (or both), which leads to a reduction in

capital costs and improved access to credit, technology and

quality inputs. As a result, more farmers are able to invest

and to increase their efficiency level. The combination of

both factors causes a reduction in the bulk of less efficient

farms and an increase in the group of (rather) efficient

farmers. Simulations based on our theoretical model are

consistent with the empirically observed changes in effi-

ciency distributions during transition.

In combination, these findings lend support to the

hypotheses that farm productivity increases are strongly

constrained by factor market imperfections, including labor

market imperfections, and limited opportunities for off-

farm employment. Farm productivity is strongly correlated

with general institutional reforms and reforms of the sec-

tors ‘‘surrounding agriculture’’.

6 Technology, resource endowments, and the nature

of productivity gains

In our conceptual framework in Sect. 2 we have outlined

that one should expect a complex relationship between

resource endowments, technological change and produc-

tivity change. The evidence on productivity changes that

we present in the rest of this paper indicates that there

exists indeed a link between technology, policy and pro-

ductivity changes. In terms of policy effects, our evidence

is indeed consistent with the hypotheses that substantial

progress in reforms is required for productivity growth to

occur. That said, the productivity impact of specific policy

reforms differs among countries.

For example, although gains in productivity have come

both from property rights reforms and organizational

restructuring, the relative importance of each component

differs between countries reflecting technology and policy

differences. In countries with labor-intensive technologies

(such as Albania) the shift from large-scale collective

farming to small-scale individual farming caused dramatic

gains in technical efficiency with relatively small losses in

scale efficiency. In capital and land intensive regions (such

as Czech Republic and Slovakia), gains in labor produc-

tivity, if any, came primarily from large farms shedding

labor with privatization of the farms (see Table 7 for an

overview of the labor intensities at the start of transition).

These different sources of productivity gains are not

coincidental. Technology and a region’s resource endow-

ments have an important impact on the relative efficiency

of different farm organizations, and thus on the incentives

for farm restructuring. As we already indicated in Sect. 2,

technology affects both the costs and benefits of the shift to

individual farming. An important factor in the optimal

scale of farming is transaction costs in labor management.

Large operations in agriculture face transaction costs

because of principal agent problems and monitoring costs

in labor contracting, which are typically large in agriculture

(Pollak 1985). Hence, individual farming will improve

labor effort and a farmer’s control over farm activities and

this will lead to efficiency gains. However, the importance

of these efficiency gains vary with specialization and

technology (Allen and Lueck 1998). Since the greatest

improvement in efficiency from farm individualization is

attributable to rising effort from better incentives, the

benefits will be relatively greater for systems in which

labor plays a greater role.

10 Empirical evidence suggests that changes in the relative wages and

in access to unemployment benefits or pensions has a significant

impact on labor use in agriculture during transition (Dries and

Swinnen 2002; Swinnen et al. 2005).

254 J Prod Anal (2010) 33:241–258

123

However, there are also costs that are incurred when

collective or corporate farms are broken up into individual

farms. In many cases there are two major types of costs. First,

there is one set of costs that could arise due to the loss in scale

economies. As in the case of the incentive effects, the impact

on scale economies will be sensitive to the nature of the

technology. The economy of scale losses may be consider-

able in the case of capital intensive production systems,

systems in which we would expect economies of scale to be

relatively significant since there are many fixed expenses and

many large assets used in farming activities. In countries in

which farming is labor intensive and few capital inputs are

used, however, such losses could be minimal.

Second, there also may be costs associated with disor-

ganization that will occur with the restructuring of farms.

The costs will arise from the mismatch that can occur

between the farm’s needs for inputs, services and equip-

ment and the infrastructure that has been set up to provide

those inputs and services. Initially designed for large scale

farming, the inputs and services that the nation’s agricul-

tural input supply chain are set up to provide are not always

suitable for individual farms. Hence, newly formed indi-

vidual farms may require an entirely different set of inputs,

services and equipment. The disorganization and econo-

mies of scale costs could be high (initially) if such inputs,

services, and equipment play an important role in the local

farming systems. Again, this is affected by technology.

These disruption costs are more likely to be lower in labor

intensive systems than in more advanced, integrated and

capital-intensive agricultural systems.

The importance of technology (resource intensity) in the

growth of individual farming is illustrated empirically by

Table 7 Selected initial

condition indicators and land

reform procedure in ECA

Source: Macours and Swinnen

(2002) and Swinnen and

Rozelle (2006)a Pre-reform indicators are

from 1989 for the Central

Europe and the Balkans, from

1990 for Asia, Transcaucasus

and the Baltics

Initial conditionsa Land reform

procedureShare of agriculture

in empl (%)

Labor/land

(person/ha)

Agricultural land

in ind. farms

Central Asia

Mongolia 32.7 0.002 0 Distribution

Kazakhstan 22.6 0.008 0 Distribution

Kyrgyzstan 32.6 0.054 4 Distribution

Tajikistan 43.0 0.185 4 Distribution

Turkmenistan 41.8 0.015 2 Distribution

Uzbekistan 39.2 0.109 5 Distribution

Transcaucasus

Armenia 17.4 0.218 7 Distribution

Azerbaijan 30.7 0.203 2 Distribution

Georgia 25.2 0.217 12 Distribution

European CIS

Belarus 19.1 0.105 7 Distribution

Moldova 32.5 0.269 7 Distribution

Russia 12.9 0.044 2 Distribution

Ukraine 19.5 0.118 6 Distribution

Baltics

Estonia 12.0 0.072 4 Restitution

Latvia 15.5 0.085 4 Restitution

Lithuania 18.6 0.098 9 Restitution

Central Europe

Czech 9.9 0.122 1 Restitution

Hungary 17.9 0.131 13 Restit ? distr ? voucher

Poland 26.4 0.258 76 –

Slovakia 12.2 0.139 2 Restitution

Balkans

Albania 49.4 0.627 3 Distribution

Bulgaria 18.1 0.132 14 Restitution

Romania 28.2 0.204 14 Restitution ? distr.

Slovenia 11.8 0.116 83 –

J Prod Anal (2010) 33:241–258 255

123

Fig. 4, which shows a strong positive relationship between

the pre-reform labor intensity of farming and the impor-

tance of individual farming 5 years after the start of tran-

sition. As such, the farm restructuring process, in particular

the growth of individual farming, is at least partially

endogenous in this transition process.

In countries with labor-intensive technologies there is a

strong shift from large-scale collective farming to small-

scale individual farming and with it strong gains in technical

efficiency with relatively small losses in scale efficiency, as

we documented above. For example, in countries such as

China, Vietnam, Albania, Armenia, Georgia, and Romania,

significant gains in productivity came mostly from the shift

to household farming when land was distributed to rural

households. In all these countries the labor/land ratio was

over 0.2 persons per hectare and total factor productivity

increased strongly during early transition (between 4 and 9%

annually) when individual farming grew from 8% of total

land use on average to 84% on average (Table 7).11

In contrast, in capital and land intensive regions, large-

scale corporate farming remained important and produc-

tivity gains came primarily from large farms shedding

labor with privatization of the farms. For example in the

Czech Republic, Slovakia and Hungary, countries in which

farming was more capital and land intensive (man/land

ratio of 0.14 or less), gains in labor productivity came

primarily from large farms shedding labor with privatiza-

tion of the farms. During the first 5 years of transition,

labor use declined by 44% on average in these three

countries (Table 4), yielding an annual increase in labor

productivity of 7.5% on average, while individual farms

used only 15% of the land (Table 1).

Notice that these differential paths not only occurred

across countries, but also across different regions within

the same country. For example, in Kazakhstan, the northern

extensive grain production systems have witnessed very

different adjustment paths, with a continuation of large

scale farming, than the southern regions where there has

been a much more radical shift towards labor intensive

household-based farming. Another example is from

Poland, where the southern and eastern regions were

characterized by small scale relatively labor intensive

production, while the western and north-western regions

were characterized by large scale state farms. In the latter

region, there has been a rapid outflow of labor from agri-

culture in the early transition years (mimicking the process

in Czech, Slovakia and Hungary) with strong increases in

labor productivity, while in the eastern and southern

regions, there has been much less of such adjustments, with

surplus labor remaining on the farms (see Fig. 5 for an

illustration of this differential adjustment process). As a

result there has been a divergence instead of convergence

in terms of farm technology and factor intensity in the

different Polish regions during the first decade of transition.

A final note relates again to the interaction with reform

policies in this process of productivity growth. Productivity

gains on small farms in labor intensive regions have

emerged only after land has been distributed to households,

a process which occurred early on in transition in countries

such as Armenia and Albania, but considerably later in, for

example, Azerbaijan, Kyrgyz, and Moldova. Yield and

productivity gains have increased substantially (or taken

off) only after these changes in land policy.

7 Concluding remarks

There have been dramatic changes in productivity over the

past 15 years in transition countries. In general one

Balkan

Caucasus

Central EurCore CIS

China

0

20

40

60

80

100

0.40 0.2 0.6 0.8

Pre-reform labor intensity (person/ha)

Indi

vidu

al fa

rmin

g 5

year

s af

ter

refo

rm (

shar

e of

land

use

)

Fig. 4 Labor intensity and the shift to individual farming. SourceSwinnen and Rozelle (2006)

NW

No

SW

NECW So

Ce WaCE

SE

-50

-40

-30

-20

-10

0

10

30 40 50 60 70 80 90 100

Share of land in individual farms (1990)

Cha

nge

in a

gr. l

abou

r (9

0-97

)

Fig. 5 Initial farm structure and labor adjustment in Poland. SourceDries and Swinnen (2002)

11 A further argument that can be made on this, pushing the

endogeneity argument even further, is that in labor intensive

economies, ultimately a land reform process emerged that was

conducive to farm individualization. This land reform procedure, i.e.

distribution of land in kind to households, often came about only after

changes were made to the existing policies, such as in Azerbaijan,

Kyrgyz Republic, Moldova, etc., reflecting changes in governments

and political economy pressures (Swinnen and Rozelle 2006).

256 J Prod Anal (2010) 33:241–258

123

observes a J-shaped (or U-shaped) effect: an initial decline

in productivity and a recovery later on. Virtually all

countries witnessed an initial decline in productivity, and

virtually all countries currently witness an increase in

productivity, and in several transition countries the growth

in productivity over the past 5 years is quite spectacular.

However, the depth and length of the initial decline

differs enormously between countries. Our analysis suggest

that the productivity changes were related to the extent of

the pre-reform distortions and the reform implementation

in the countries. In the most advanced reformers (mostly in

Central Europe) the decline was relatively mild and

recovery started fairly early on, and productivity growth, in

particular labor productivity, has been strong for more than

a decade. In the Baltics, recovery started relatively quickly,

reflecting the fast pace of the reforms, but the initial fall in

productivity was much deeper than in Central Europe,

possibly reflecting the larger pre-reform distortions as these

countries were part of the FSU. In many of the CIS

countries, the decline of productivity was much more

dramatic (around 25% in several countries) and lasted for

much of the 1990s. It was only after the Russian crisis that

a recovery started, and although hesitant at first, produc-

tivity increases are strong, certainly for yields.

However, early recovery and growth was not limited to

the richer Central European countries. Remarkably, a

country such as Albania has recorded strong productivity

growth continuously since the start of the land and farm

reforms in 1992. The combination of immediate Chinese-

style land reform giving land rights to rural households in a

very labor intensive production system and widespread

migration with remittances which have provided access to

finance for the remaining farm population have allowed

productivity growth to start early and continue since.12

It is clear that policies have played a very important role

in this process. Those that have implemented the necessary

reforms early on, such as the Central European countries,

have started to witness productivity growth earlier. How-

ever it is also clear that initial conditions mattered. The

institutional and human capital hurdles to create a market

economy were larger in the countries further east and

therefore it is no surprise that the transition productivity

declines have been deeper and longer.

What is remarkable is that the nature of productivity

gains has differed starkly between countries, and even

between countries fairly close in other aspects of develop-

ment and reforms. In countries such as Hungary, Czech

Republic and Slovakia, there has been very strong labor

productivity growth from the outset of transition. In these

countries, large private farming companies have survived

and have laid off many workers, thereby driving labor

productivity to grow. This is very different from the reforms

in, for example, Poland, another Central European country.

There, such process only took place in the North and the

Western parts of the country with the privatization of the

large state farms. In other parts where small farms domi-

nated, there was no reduction of labor for the most part of

the 1990s and productivity gains have been more limited.

The productivity growth in labor intensive countries

such as Albania, Azerbaijan, etc has not come from laying

off workers, since in many cases, there has not been a

significant outflow of labor from agriculture, at least not in

the first decade of transition. Initially it came from land

reforms, providing better incentives for labor governance

in labor intensive agricultural systems. Once this one-off

productivity shock from land reform (which typically

needs to come through the distribution of physical plots of

land to rural households) is ‘‘consumed’’ it is crucial to

focus on other sources of productivity growth.

More recent growth in productivity has come from a

combination of improved options for employment outside

of agriculture (either in other sectors or through migration),

from improvements in social payments (pensions and

unemployment benefits) allowing subsistence farmers to

leave agriculture, and from improved access to factor and

output markets. The latter has come about through general

reforms which have stimulated investments in the food

industry (including the inflow of foreign investment) which

brought new technology and capital into the agrifood

chains, and a more stable general investment climate

leading to off-farm employment creation.

Further the improvement of competitiveness with the

ruble devaluation and of the fiscal situation in Russia and

some of the other energy and mineral exporting with

increasing international prices have led to higher demand,

more timely payments, higher farm profits, etc, all of which

have improved the financial situation of the farms and

thereby their opportunities to use better inputs and to invest

in new technology in these countries, and to less disrup-

tions in the exchanges between farms and the up- and

downstream companies. All these factors have contributed

to increased productivity.

Acknowledgments The authors thank Anneleen Vandeplas for

research assistance and Azeta Cungu for comments and providing

access to data and estimation procedures. The views in this paper are

those of the authors and not necessarily reflect those of organizations

they are associated with or those that funded the research.

References

Allen DW, Lueck D (1998) The nature of the farm. J Law Econ

41:343–386

12 See Cungu and Swinnen (1998) on the special character of the

Albanian reform process.

J Prod Anal (2010) 33:241–258 257

123

Arnade C, Gopinath M (1998) Capital adjustment in US agriculture

and food processing: a cross-sectoral model. J Agric Resour

Econ 23(1):85–98

Bofinger P (1993) The output decline in Central and Eastern Europe:

a classical explanation. CEPR discussion paper, London

Brada JC (1989) Technical progress and factor utilization in eastern

european economic growth. Economica 56:433–448

Charnes A, Cooper WW, Rhodes E (1978) Measuring the Efficiency

of Decision Making Units. Eur J Oper Res 2:429–444

Cungu A, Swinnen J (2003) Transition and total factor productivity in

agriculture 1992–1999. PRG working paper

Dries L, Swinnen J (2002) Institutional reform and labor reallocation

during transition: theory and evidence from polish agriculture.

World Dev 30(3):457–474

Dries L, Swinnen J (2004) Foreign direct investment, vertical

integration and local suppliers: evidence from the polish dairy

sector. World Dev 32(9):1525–1544

Dudwick N, Fock K, Sedik D (2005) A stock-taking of land reform

and farm restructuring in Bulgaria, Moldova, Azerbaijan and

Kazakhstan. World Bank report

FAO (1986) Inter-country comparisons of agricultural production

aggregates. FAO Economic and Social Development paper no.

61, Rome

Fare R, Grosskopf S, Lovell C (1985) The measurement of efficiency

in production. Kluwer, Boston

Gow H, Swinnen J (1998) Up- and downstream restructuring, foreign

direct investment, and hold-up problems in agricultural transi-

tion. Eur Rev Agric Econ 25(3):331–350

Hyami Y, Ruttan VW (1970) Factor Prices and technical change in

agricultural development: The United States and Japan, 1880–

1960. J Polit Econ 78(5):1115–1141

Hyami Y, Ruttan VW (1985) Agricultural development: an interna-

tional perspective. Johns Hopkins University Press, Baltimore,

MD

International Labour Organization (ILO) (1997–2002) Yearbook on

labour statistics. ILO, Geneva

Jackman, R (1994) Economic policy and employment in the transition

economies of Central and Eastern Europe: what have we

learned? Int Labour Rev 133(3):327–345

Koppel B (1995) Induced innovation theory and international

agricultural development: a reassessment. Johns Hopkins Uni-

versity Press, Baltimore, MD

Lerman Z (2008) Agricultural recovery in the former Soviet Union:

an overview of 15 years of land reform and farm restructuring.

Post-Communist Econ 20(4):391–412

Lerman Z, Csaki C, Feder G (2004) Agriculture in transition: land

policies and evolving farm structures in post-Soviet Countries.

Lexington Books, Lanham, MD

Liefert M, Gardner B, Serova E (2003) Allocative efficiency in

Russian agriculture: the case of fertilizer and grain. Am J Agric

Econ 85(5):1228–1233

Macours K, Swinnen J (2000a) Causes of output decline during

transition: the case of Central and Eastern European agriculture.

J Comp Econ 28:172–206

Macours K, Swinnen J (2000b) Impact of initial conditions and

reform policies on agricultural performance in Central and

Eastern Europe, the former Soviet Union, and East Asia. Am J

Agric Econ 82(5):1149–1155

Macours K, Swinnen J (2002) Patterns of Agrarian Transition. Econ

Dev Cult Change 50(2):365–395

Mathijs E, Swinnen J (2001) Production efficiency and organization

during transition: an empirical analysis of East German agricul-

ture. Rev Econ Stat 83(1):100–107

Petrick M (2004) Farm investment, credit rationing, and governmen-

tally promoted credit access in Poland: a cross-sectional analysis.

Food Policy 29(3):275–294

Pollak RA (1985) A transaction cost approach to families and

households. J Econ Lit 23(2):581–608

Rozelle S, Swinnen J (2004) Success and failure of reforms: insights

from transition agriculture. J Econ Lit 42(2):405–456

Sedik D, Trueblood M, Arnade C (1999) Corporate farm performance

in Russia, 1991–1995: an efficiency analysis. J Comp Econ

27:514–533

Seeth HT, Chachnov S, Surinov A, von Braun J (1998) Russian

poverty: muddling though economic transition with garden plots.

World Dev 26(9):1611–1623

Swinnen J, Rozelle S (2006) From Marx and Mao to the market. The

economics and politics of agricultural transition. Oxford Uni-

versity Press, Oxford

Swinnen J, Vranken L (2005) Reforms and efficiency changes in

transition agriculture. LICOS discussion paper no. 172

Swinnen J, Dries L, Macours K (2005) Transition and agricultural

labour. Agric Econ 32(1):15–34

258 J Prod Anal (2010) 33:241–258

123