A

GM

a

ARRA

KLFLLTCR

I

nihitTu

n2pwertuoaift

0d

Land Use Policy 26 (2009) 1080–1089

Contents lists available at ScienceDirect

Land Use Policy

journa l homepage: www.e lsev ier .com/ locate / landusepol

ssessing land reallocation decisions during transition in Romania

eorgeta Vidican ∗

asdar Institute of Science and Technology, P.O. Box 54224, Abu Dhabi 02141, United Arab Emirates

r t i c l e i n f o

rticle history:eceived 8 August 2008eceived in revised form 10 January 2009ccepted 15 January 2009

eywords:and reform

a b s t r a c t

The post-socialist countries underwent dramatic changes in agricultural land ownership and produc-tion system. Former socialist collective farms have been dismantled, rural land has been privatized,and land markets have been formalized. Nevertheless, more than a decade later farming associationsstill persist in the choices that landowners make in terms of land reallocation despite collective actionproblems and the availability of leasing-out land as a close substitute. While the decision of farmingthe land individually has been well understood, there is less research on why landowners join farm-

arming associationsand marketsand leasingransition countriesentral and Eastern Europe

ing associations rather than participate in land transactions. The paper examines this question usinghousehold survey data from the two largest agro-regions in Romania. I find that farming associations con-stitute a good land reallocation option for landowners that are resource constrained but are still willingand able to be engaged in farming. Associations provide security of tenure and capital access, allowinglandowners to draw on the benefits of scale economies. Leasing-out is a viable alternative for younger

age in

ofciafatlS

a(g(

ct(

omania landowners who can engabilities.

ntroduction

The fall of the Berlin Wall, symbolizing the demise of commu-ism, ushered a process of change unmatched in its scale and speed

n economic and political history. The first and the most compre-ensive institutional reforms in the early 1990s were implemented

n the agricultural sector, signaling a major ideological change—theransfer of property rights in land from the state to the individuals.he social and economic changes generated by land reform werenprecedented in complexity and scale.

In Romania, the post-socialist land reform marked the begin-ing of a series of transformations in ownership structure. By005 almost 95% of land was used in individual farms as com-ared to only 9% in 1990 at the start of transition. Collective farmsere dismantled, land was successfully restituted to former own-

rs based on the 1940s land records, and state farms were slowlyeorganized into large corporate farms. Almost overnight morehan four million new landowners were created, establishing annprecedented level of private property. However, the slow pacef implementing institutional reforms for the newly created small

nd medium farms resulted in extreme land fragmentation, lim-ted access to markets and, in effect, strong challenges to transitionrom subsistence farming to more commercially oriented produc-ion.

∗ Tel.: +971 50 328 5263.E-mail address: gvidican@mist.ac.ae.

fmom2

st

264-8377/$ – see front matter © 2009 Elsevier Ltd. All rights reserved.oi:10.1016/j.landusepol.2009.01.001

non-farming activities and for older landowners with limited farming

© 2009 Elsevier Ltd. All rights reserved.

Contrary to the expectations of transition policy makers, vari-us forms of farming associations widely persisted throughout theormer communist countries in the Central and Eastern Europeanountries and the Commonwealth of Independent States follow-ng de-collectivization (Meurs, 2001; Verdery, 2003; Lerman etl., 2004; Sabates-Wheeler, 2005; Allina-Pisano, 2007). Individualarming emerged, but in general farms have been too small to takedvantage of economies of scale (Slangen et al., 2004). At the sameime, participation of landowners in land markets has been veryimited (Vranken and Swinnen, 2006; Lerman and Shagaida, 2007;winnen and Vranken, 2007).

By 1993 more than 40% of land was voluntary returned to associ-tions in Romania, despite rapid de-collectivization and land titlingBrooks and Meurs, 1994). Similar outcomes were recorded in Bul-aria, where more than 40% of land is being farmed in associationsLerman et al., 2004).

The reallocation of land in farming associations following de-ollectivization and land restitution occurred despite the fact thathis farming arrangement is widely criticized in the literatureCarter, 1984; Pollak, 1985; Ellickson, 1993) and instead individualarming is portrayed as panacea for the agricultural sector in the for-

er socialist countries (World Bank, 2007). Hence, the reluctancef rural households to engage in individual farming was often dis-

issed as irrational, and ideologically motivated behavior (Meurs,

001).The persistence of associations is intriguing also because in the

econd half of the 1990s the opening up of land markets made wayo alternative farming arrangements, such as leasing-out and sales.

olicy

Lclfsu

tdi(to

lclaegfl

etoanjlaptIlessrco(ni

I

frtb1iptitabAacw

tbwr1rDa2t

(itrba

f

TT

T

FNAA

INAA

PNAA

nF

(

G. Vidican / Land Use P

and market institutionalization was viewed as the solution to theurrent level of low agricultural performance and the high degree ofand fragmentation. As a result, policy makers assumed that smallarmers would engage in land market transactions in order to con-olidate scattered plots and to release assets to more productivesers (Duncan and Prosterman, 2000).

The goal of this study is to explain the persistence of associationshroughout the transition period. I examine factors associated withifferent land reallocation decisions in the two main agro-regions

n Romania, the Western Plain (WP) and the Central Romanian PlainCRP). Agro-regions are clusters of counties that have similar pat-erns of agricultural land use (arable land, pastures, vineyards, andrchards) (Sandu, 1999, p. 17).

The research question that shapes the analysis is: Why doandowners persist in joining associations in Romania, despite per-eived collective action problems, and despite the availability ofeasing-out as a close alternative for land reallocation? To fullynswer this question, two accompanying questions need to bexplored: (a) Why do households seek alternative farming strate-ies instead of farming the land individually? (b) What are theactors that affect landowners’ choices between associations andeasing-out arrangements?

While some of these aspects were previously examined (Rizovt al., 2001; Sabates-Wheeler, 2005), there is limited research onhe factors affecting the choice between associations and leasing-ut, primarily because of data limitations. This analysis is based onhousehold survey in the two largest agricultural regions in Roma-ia, conducted in 2006. I find that one of the main benefits from

oining farming associations resides in the ability to consolidateand plots while maintaining full ownership of the land. The later isn important aspect derived from the historical legacies of privateroperty rights prior to collectivization, as suggested by the “collec-ive memory” argument (Hann, 1993; Vidican, 2008). In addition,find that leasing-out is a preferred farming alternative only for

andowners that are older, capital constrained, and for landown-rs with non-farming income sources. This suggests that the largerocio-economic environment is critical for land reallocation deci-ions. Moreover, the results from this study show that in deciding toeallocate land between different farming arrangements the con-

epts of costs and benefits are much broader than at the first glancene may think. Hence, the findings confirm earlier explanationsRizov et al., 2001; Sabates-Wheeler, 2005) and add additionaluances to understanding why some households seek alternative

nstitutional arrangements rather than farm all land individually.

etevS

able 1he pattern of land reallocation at the national level between 1993 and 2005.

1993

otal arable land privatelya farmed (1000 ha) 11,006

ormal associationsumber 4,266gricultural land (% from total arable land) 17.4verage size (ha) 448

nformal associationsumber 13,772gricultural land (% from total arable land) 16.0verage size (ha) 128

rivate individual farmsumber (100 ha) 3,420gricultural land (% from total arable land) 66.6verage size per farmer (ha) 2.1

.a. = data not available from official statistics. Source: For 1993–1999, National Program food Newsletter (July 2000) in Aligica and Dabu (2003). For 2002, NIS (2002); for 2005, M

a Private refers to land that is in private ownership (not in state or cooperative ownersthrough leasing).

26 (2009) 1080–1089 1081

nstitutional arrangements for farming

Romania was the last country in Eastern Europe to break awayrom its post-war era and to start implementing the transitioneform programs in the 1990s. Since then, the structure and insti-utions that were established by the communist government haveeen dismantled. The post-socialist land reform, implemented in991, was unprecedented in the degree of institutional restructur-ng that took place. Private property rights were re-instated to there-1940s levels, former collective farms were dissolved or restruc-ured into smaller formal associations, new endogenously formednformal associations were created, state farms were slowly priva-ized, and the former state owned channels for product collectionnd marketing, as well as the networks for input distribution, wereroken down and in some cases private intermediaries surfaced.ll these changes resulted in significant challenges for farmers andgricultural professionals alike. At least until mid-1990s, the out-ome was a real “muddling through” in an environment marked byide uncertainties in a constantly changing legal system.

The alternatives for land reallocation varied significantly overhe years. Immediately after land reform farmers could chooseetween farming individually or in different forms of associations,hile land markets were formalized only later. Table 1 shows land

eallocation between different institutional arrangements between993 and 2005. These institutional alternatives can be viewed as aearrangement of property rights over a spectrum of tenure forms.espite the fact that the share of land farmed in associations (formalnd informal) declined since earlier in the transition period, still by005, at least 5% of the land in private ownership is managed underhis arrangement.

Such outcome is not unique to Romania. For example, Stark1996), points to similar patterns in Hungary, Lerman et al. (1998)n Moldova, and Lerman et al. (2007) in Ukraine. In describinghese arrangements Verdery (1999) used the term “fuzzy propertyights” to portray the wide variation of property relations, which arelurred and ambiguous in terms of obligations and responsibilitiesssociated with the restituted land.

Following restitution and the transfer of property y rightsrom the state to individuals, an increasing number of landown-

rs decided to farm the land themselves. Earlier research showedhat households that are younger, more educated, and have a higherndowment with physical capital are more likely to engage in indi-idual farming (Mathijs and Swinnen, 2001; Rizov et al., 2001;abates-Wheeler, 2005). In this analysis, individual farmers refer

1996 1999 2002 2005

11,539 11,921 13,931 13,907

3,759 3,573 2224 161415.2 11.9 7.0 5.3466 396 432 455

15,107 6,264 n.a. n.a.12.5 7.3 n.a. n.a.95 139 n.a. n.a.

3,626 4,120 4,277 4,10372.3 78.7 55.3 65.42.3 2.3 1.7 2.2

or Agriculture and Rural Development 1999, Annex 34, Ministry of Agriculture andARD (2007).hip), including land farmed by private individual farms, associations, and tenants

1 olicy 26 (2009) 1080–1089

tpuahtutm

tlmttMiovtmaaef

amnvtoa

p(Ttacafeoelmti(a(obtp

M

S

ida

Table 2Key descriptive statistics from the 2006 household survey in Romania.

Western Plain Central Romania Plain

Household size (# persons) 3.3 3.00Age of the household head (years) 57 64Gender of the household head (% male) 57.0 63.0Education of the household head (%

who completed 8 years)63.0% 79.0

Main occupation (% in agriculture) 52.0 86.0Secondary occupation (% in

agriculture)25.0 6.0

Occupational status (% pensioners) 55.0 69.0Average landownership (ha) 4.0 4.0Land fragmentation (# of plots) 3.8 4.0Final ownership title (%) 86.7 93.3Joined farming associations (% of

households)4.1 35.6

Leased-out land (% of households) 10.4 29.7Leased-in land (% of households) 4.4 2.3Sold land (% of households) 22.2 3.3Bought land (% of households) 18.4 4.0Own tractor (% of households) 21.1 4.3Sold agricultural products (% of

households)40.2 24.4

S

SR

om

h(htRfal

s(vittatws12s

aca

foTable 3 shows the distribution of households in the choice cat-egories for the two dependent variables. Because the dependentvariables are binary, the analysis of data is based on probit regres-

082 G. Vidican / Land Use P

o landowners with a final land title as well as those with a tem-orary title (“adeverinta de punere in posesie”) who farm the landsing household and/or hired labor and who are not engaged incontractual arrangement (formal or informal) with other house-olds or with agricultural entrepreneurs for farming the land. Thisype of farming assumes a complete bundle of rights over the landnder cultivation and over the produce derived from the land (withhe exception of land that was leased-in under a contractual agree-

ent).Immediately after restitution and the breakdown of the collec-

ive farms, associations emerged as alternatives to farming all theand individually. Associations were of two types: formal and infor-

al. Formal associations were created directly on the structures ofhe former collective farms (i.e. same buildings, capital, and often-imes the same management) although significantly smaller in size.

embership in formal associations is easy to obtain and final titles not required for joining. The rent is generally paid as a sharef the harvest at the end of the year (on average 30% of the har-est). Nevertheless, a specific characteristic of the associations ishat members contribute with labor to farming activities, while

echanical work for land preparation and harvesting is done usingssociations’ agricultural equipment. A major advantage of formalssociations, as opposed to other farming arrangements, is the ben-fit members derive from the networks of production inheritedrom the old collective farms.

Informal associations were created endogenously by neighborsnd relatives in order to consolidate land plots and to rely on com-on resources (i.e. capital, labor). These farming arrangements do

ot have legal status, restricting members’ access to financial ser-ices. Because, informal associations declined significantly sincehe beginning of land reform (see Table 1), this paper examinesnly the formal associations (see Sabates-Wheeler (2005) for annalysis of the informal associations).

Land market transactions (i.e. leasing-in, leasing-out, sales andurchases), as another form of alternative farming arrangementsor tenancy), were formalized only later in the transition process.he first leasing law was passed in 1994 (Law 16/1994), addressinghe visible lack of viable frameworks for basic economic trans-ctions in agriculture, and was viewed as an alternative for landonsolidation in larger farming units (Aligica and Dabu, 2003). Inddition, the Law sought to find ways to transfer land use rightsrom farmers that could not work all the land in ownership to farm-rs that had the capacity (labor and capital) to enlarge their farmperations but were unwilling (or unable) to purchase land. How-ver, because of the heavy restrictions imposed on the eligibility toease-in or out (related to land registration and contractual arrange-

ents) and on the nature of leasing arrangements (the duration ofhe lease was set to at least 5 years, and a minimum rent was spec-fied), the leasing law was slow in promoting a fluid leasing marketDobrescu, 1999; Sabates-Wheeler, 2005). The rent for the leasingrrangements is generally paid as a fixed share of the yearly harveston average the equivalent of 600 kg of wheat). In Romania this formf tenancy is different from the canonical model of sharecroppingecause the tenants are generally capital rich, having easier accesso financial resources. Land sales were institutionalized in 1998 buturchases and sales of land have been very limited.

aterials and methods

ources of data

For this analysis I use household survey data from a cross-sectionn the two largest agro-regions in Romania for the year 2006. Thisata builds on a national household survey implemented a decadego. The first was a national survey conducted in 1996 on a sample

s

t

ample size (N) 316 303

ource: Data from the 2006 household survey in the Western Plain (WP) and Centralomanian Plain (CRP).

f 1650 households with joint funding from the Romanian Govern-ent, the European Commission and the World Bank.In the summer of 2006 I surveyed 619 randomly selected house-

olds owning at least half a hectare of agricultural land in 45 villagessame villages as in the 1996 survey) in two agro-regions: 316ouseholds in the Western Plain (WP) and 303 households in Cen-ral Romanian Plain (CRP). The two regions cover more than half ofomania’s arable land and have historically been the breadbasketsor the country. Nevertheless, the economic profiles of WP and CRPre different, with the west region being wealthier and having aonger history of private property rights (Vidican, 2008).

For selecting the households to be surveyed I used the sameampling strategy as in 1996 and the same research instrumentwith a few additional questions and slight changes in the earlierersion). Hence, the two surveys represent repeated cross-sectionsn the same villages, reason for which I cannot draw conclusions onhe policy driven changes between 1996 and 2006. Nevertheless,he availability of this data allows me to trace changes in the socialnd economic conditions of farmers in the selected villages, ando evaluate decisions on land reallocation over a period of timehen significant institutional changes were implemented. Table 2

hows descriptive statistics for the household surveys. Because the996 data has been analyzed elsewhere (Rizov et al., 2001; Rizov,003), I focus here primarily on the results from the 2006 householdurvey.1

In addition to household surveys, I conducted a survey for thegricultural specialists in the communes (or the mayor in someases) to gather information at commune level on land use, labornd land markets, and institutional constraints for private farmers.

The dependent variables for the analysis are: (1) the choice ofarming all land individually vs. not (i.e. in associations or leasing-ut ) and (2) the choice of joining associations vs. leasing-out land.

ion models.

1 Vidican (2008) offers a larger discussion on the comparative analysis betweenhe two regions based on both survey data.

G. Vidican / Land Use Policy

Table 3Summary stats for the dependent variables used in the discrete choice model (2006).

Percentage

First choice levelFarming all land individually 64.5Choosing alternative arrangements 35.5

Second choice levelJoining associations 43.1L

SR

vah

R

ratda

U

eaadi

U

c

ictdau

V

wt

miitaw

P

vmMtd

d

TS

HAE

L

A

FLT

H

L

T

S

L

CLDS

SP

easing-out 46.9

ource: Data from the 2006 household survey in the Western Plain (WP) and Centralomanian Plain (CRP).

The independent variables include a series of household specificariables farming conditions variables, commune specific variables,nd regional fixed effects (see Table 4 for summary statistics for theousehold surveys in both 1996 and 2006).

esearch methods

I use discrete choice models (probit regression) to examine landeallocation in different farming arrangements (individual farming,ssociations, leasing-out). The individual chooses the alternativehat yields the highest level of utility. Utility, in this case, can beefined as a function of returns (�), commune characteristics (�),nd social factors (�).

ij = Uij(�ij, �ij, �ij) (1)

I assign a utility level Uij to each alternative j = 1, . . ., J, forach decision maker i = 1, . . ., I. The utilities are determined bylarge number of characteristics of the decision maker and the

lternatives. The researchers have information on some of thoseeterminants, but not on all. This is reflected in splitting the utility

nto a deterministic part Vij and a stochastic part εij:

ij = Vij + εij (2)

Based on the utility maximization assumption, an individualompares U1, U2, . . ., UJ, and chooses the one with maximum utility,

efaat

able 4ummary statistics for the variables included in the analysis.

ousehold characteristicsge Age of the household head (years)ducation Years of completed education for the household

years)and–adult ratio The share of owned land to the number of adults

(ha/person)griculture main occupation Dummy variable for whether agriculture is the p

(1, agriculture primary occupation; 0, otherwise)arming conditionsand fragmentation Number of plots normalized by total land in ownitle Dummy variable for possession of ownership titl

title; 0, does not have title)ousehold labor The number of labor-days spent on the farm by h

members (days)oan dummy Dummy variable for contracting bank loans in th

(1-contracted loan; 0-did not contract loan)ractor dummy Dummy variable for whether the household own

owns; 0, it does not own)eeds dummy Dummy variable for purchasing seeds in the last

seeds; 0, did not purchase seeds)ivestock Livestock index measured in standard heads

ommune characteristicsand quality The share of quality I and II agricultural land in tistance to markets Distance from the commune to the closest city (khare of arable land Share of arable land in total land in the commun

.D. = standard deviation. The values in parentheses are standard deviations. DV = dummylain (WP) and Central Romanian Plain (CRP).

26 (2009) 1080–1089 1083

.e. max(UJ). The term Vi is often referred to as the “representativeomponent of the utility” (Hensher, 2005, p. 76) because it includeshe observed and measured attributes. In its simplest form, Vi can beefined as a linear expression in which each variable is weighted byunique weight (parameter) to account for that variable’s marginaltility input such as:

i = ˇ0i + ˇ1i(X1i) + ˇ2i(X2i) + · · · + ˇki(Xki), k = 1, . . . , K (3)

here ˇki is the parameter associated with variable Xk and alterna-ive i.

Because the researcher has less information than the decisionaker one can explain an individual’s choice only up to a probabil-

ty of an alternative being chosen. The probability Pij that individualchooses some alternative j is equal to the probability of Uij beinghe largest of all Ui1, Ui2, . . ., Uij. With yi ∈ {1, . . ., J} denoting thelternative that decision maker i chooses, this probability can beritten as (Heiss, 2002):

ij = Pr(yi = j) = Pr(Uij ≥ Uik ∀ k = 1, . . . , J; k /= j)

= Pr((Vij + εij) ≥ (Vik + εik) ∀ k = 1, . . . , J; k /= j)

= Pr((εik − εij) ≤ (Vij − Vik) ∀ k = 1, . . . , J; k /= j) (4)

This equation states that due to lack of full information, the indi-idual decision marker’s utility maximization is a random utilityaximization. The random utility approach was first formalized byanski (1977). Given the deterministic parts of the utility func-

ions Vi1, . . ., ViJ, this probability depends on assumptions on theistribution of the stochastic error terms, εii, . . ., εiJ.

The decision a landowner takes on land reallocation can beescribed by two phases: (i) given resource endowment, other

ndogenous (specific to the household) and exogenous factors, thearmers decide whether to farm all the land themselves or to seeklternative farming arrangements; (ii) if the farmers decide to seeklternative farming arrangements (i.e. conditional on this choice),hey choose between joining associations or leasing-out land.

Hypotheses Mean S.D.

+ 62 (14)head (number of − 7.53 (3.42)

in the household + 1.55 (1.99)

rimary occupation + 0.70 (0.46)

ership (plots/ha) + 1.55 (1.07)e on land (1, have − 0.90 (0.30)

ousehold + 362 (252)

e past year + 0.13 (0.34)

s a tractor (1, + 0.03 (0.17)

year (1, purchaed + 0.59 (0.49)

− 1.89 (3.80)

he commune + 43.96 (33.54)m) − 18.28 (7.46)

e (%) − 70.85 (15.11)

variable. Source: Calculations based on the 2006 household survey in the Western

1 olicy

biam

�

wsta

R

ilfalmh

H

taBtaa

ofwicuatNb

bofAl

ropl(nftnfsw

gsl

(Mifl(cmfa2

pafohthf

F

tltiipstnstt

ptaLwahtfian

lfwtnmT“

dh

084 G. Vidican / Land Use P

Hence, I estimate the choice probabilities based on two pro-it models: (a) the probability that a landowner farms all land

ndividually; (b) the probability that a landowner chooses to joinssociations, given that she opted for alternative farming arrange-ents. The regression model I am estimating is:

(Pij)=ˇ0i + ˇ1i(X1i) + ˇ2i(X2i) + · · · + ˇki(Xki), k=1, . . . , K (5)

here Pij is the probability associated with alternative j that deci-ion maker i chooses, � is the cumulative distribution function ofhe standard normal distribution, and ˇki is the parameter associ-ted with variable Xk and alternative i.

esearch hypotheses

Below I discuss the research hypotheses for the variablesncluded in the analysis: household characteristics (age, education,and–adult ratio, primary occupation), farming conditions (landragmentation, possession of land title, use of household labor,ccess to bank loans, physical capital ownership, seeds purchases,ivestock) and commune characteristics (land quality, distance to

arkets, share of arable land). Table 4 also summarizes the researchypotheses.

ousehold characteristicsBroadly speaking, human capital (age and education) reflects

he managerial skills and capabilities of rural households (Sahnnd Alderman, 1988; Foster and Rosenzweig, 1994; Rizov, 2003).ased on the 1996 survey data for Romania, Rizov (2003) showedhat human capital has a significant effect on the choice of farmingrrangements in Romania. Rizov finds that low skilled householdsre more likely to join associations.

The effect of age (measured as the age of the household head)n the choice of farming arrangements can be interpreted in dif-erent ways. Higher age generally implies a higher risk aversion,hile it can also suggest stronger managerial experience and abil-

ty. Hence, I hypothesize that age has a non-linear effect on thehoice of institutional arrangements. I assume that older individ-als have more farming experience and accumulated knowledge,nd therefore they are more likely to farm the land themselves, oro join associations if they lack key resources for individual farming.evertheless, at higher age levels (more than 70 years), leasing-outecomes a better alternative due to decline in farming abilities.

I measure the level of education as years of education completedy the household head. I hypothesize that education matters notnly as a proxy for managerial skills and ability to start up and run aarm but also as an indicator of access to off-farm job opportunities.s the level of education increases, so does the opportunity cost for

abor.The land–adult ratio is measured as the amount of owned and

ented-in land in hectares divided by the number of individualsver 15 years living in the household. This variable represents aroxy for farm size as well as a measure of wealth. The average

and–adult ratio in 1996 was 1.5 ha, almost the same as in 20061.6 ha per adult). The average size of land farmed individually didot change significantly in the past decade. In 1996 a household

armed on average 3.1 ha of land, while by 2006 the share increasedo only 3.3 ha. This finding is rather surprising since based on eco-omic theory, policy makers expected that once land markets are

ormalized and the leasing legislation becomes less cumbersome,mall farms will increase to a more economical size, closer to the

estern model of family farms.

A higher share of land relative to adults in the household sug-ests that landowners can benefit not only from economies ofcale, but also from increased opportunities for financing sinceand is most commonly used as collateral in transitional economies

apw

t

26 (2009) 1080–1089

Dumitru, 2002; Deininger, 2003; Deininger and Songqing, 2003).oreover, given that land is the most important asset for farm-

ng, more land relative to labor is supposed to encourage individualarming (Rizov et al., 2001; Mathijs and Swinnen, 2001). Neverthe-ess, when the type of agriculture practiced is subsistence-basedwhich is the case across most Eastern European and former Sovietountries), and markets are imperfect (preventing landowners toatch the amount of land with the availability of other assets),

armers with more land are likely to seek alternative farmingrrangements, such as farming associations (Mathijs and Swinnen,001).

The dummy variable for whether agriculture is the primary occu-ation proxies for the availability of alternative sources of incomend for whether the household is engaged in full time or part timearming. In 1996 83% of the households had agriculture as mainccupation, while in 2006 the share was 70%. A lower share ofouseholds with agriculture as the main occupation suggests thathe rural economy is more diversified. I hypothesize that house-olds with agriculture as the primary occupation are more likely to

arm individually or to join associations.

arming conditionsLand fragmentation is measured as the total number of plots that

he land in ownership is divided into, normalized by the size ofandownership (total land owned by a household) in order to cap-ure the degree of fragmentation, rather than a wealth affect. A plots not defined by what is cultivated on a piece of land, as is the casen the African context (Suri, 2007). Rather, in the CEE context, a landlot is defined by its physical boundaries, as a measure of physicaleparation of land ownership into different units. Land fragmen-ation can be measured by the following parameters: farm size,umber of plots, plot size, plot shape, spatial distribution of plots,ize distribution of plots. A common measure of land fragmenta-ion is Simmon’s Index (Bentley, 1987), which formed the basis forhe measure used in this analysis.

Land fragmentation is probably the most frequently mentionedredicament to improving performance in the Romanian agricul-ure, not only by policy makers in Romania, but also by researchersnd academics for other country case studies (Lerman, 2005;erman and Cimpoies, 2006; Lerman et al., 2007). Nevertheless,hen market imperfections are pervasive, land fragmentation can

lso act as an economic risk buffer (Blarel et al., 1992) despiteigher costs incurred by needing more labor at peak harvestingimes and higher commuting time between the different plots. Landragmentation can reduce risk through differences in land qual-ty among plots (Scott, 1976). But, despite the potential positivespects of fragmentation, the costs generated as a result are notegligible.

I hypothesize that at higher degrees of land fragmentation,andowners benefit more from farming in associations (rather thanrom leasing-out), since consolidation can be achieved easier andith lower transaction costs (costs associated with land registra-

ion and setting up contracts). To join associations, landowners areot required to have full ownership title and the membership agree-ents they sign do not require to be notarized by a local official.

he enforcement of the membership agreement is made throughpeer-pressure” in the local community.

In the past decade significant progress was made in land titleistribution. By 2006 almost 98% of the households had title deeds. Iypothesize that households with more secure property rights seek

lternative farming arrangements. This outcome should be moreronounced earlier in the transition process as opposed to laterhen property titles were more widely distributed.

I measure household labor by the number of days spent onhe farm by household members (part time and full time). I

olicy

hlfrtlt

hvtobl

stgtBto2nrimt

ovhodc

slwpita

C

gstte

fialid

csta

t

co

R

dilatt

Ca

lal8fiitr

p2iiitdhtwawr

hlmn

mibahalillbecause of high transaction costs associated with the alternativefarming strategies (leasing-out in particular), it is easier for house-holds with more land to participate in land transactions. Fig. 1reveals differences between the two agro-regions in terms of the

G. Vidican / Land Use P

ypothesize that farmers using more household labor are moreikely to farm the land themselves rather than to choose alternativearming arrangements. Labor availability is problematic in theural areas, due to high level of youth migration and the old age ofhe rural residents (Diminescu, 1996). In discussions with farmersabor constraints have been raised as one reason for choosing noto farm all the land individually.

Access to loans is measured as a dummy variable for whether theousehold contracted a bank loan in the previous year or not. Thisariable aims to capture the ability of the household to invest andake risk. In addition, access to loans also reflects the availabilityf assets for collateral. I hypothesize that farmers that contractedank loans are more entrepreneurial, asset rich, and hence more

ikely to farm the land themselves.Physical capital is very scarce in the Romanian countryside. The

urvey included a list of eight different physical capital items: truck,ractor, plough for tractor, combine for cereals, carriage, seeder, irri-ation equipment, and processing equipment. The distribution ofhese different agricultural equipments across households varies.ased on the 2006 survey data 13% of the households owned a trac-or, 12% owned plough for tractor, and fewer than 6% owned anyther items, with the exception of carriages that were owned by9% of the households. Owning physical capital allows a householdot only to farm at a larger scale, but also to derive a source ofevenue by renting services to other farmers. Since the availabil-ty of tractor is critical for improving farming productivity through

echanization, I use a tractor dummy variable to measure capi-al.

Another variable for capturing farming conditions is a measuref whether the household purchased seeds in the past year (dummyariable). Purchasing agricultural inputs suggests a tendency of theousehold to move away from traditional farming, where the usef seeds produced by individual farmers (small landowners) pre-ominates. Hence, purchasing inputs indicates both availability ofash and entrepreneurship.

Livestock is a measure of resource endowment as well as diver-ification away from crops, calculated as the sum of weightedivestock heads existent in the household. I used the following

eights: dairy cow = 1; calve = 0.1; bull = 1; horse = 1.1; pig andiglet = 0.1; sheep, lamb, and goat = 0.1; poultry = 0.01. The result-

ng livestock variable is measured in standard heads. I hypothesizehat a larger livestock suggests that a household is likely to seeklternative farming arrangements, or to lease-out the land.

ommune characteristicsLand quality is measured as the share of good quality land (cate-

ories I and II) in the commune level agricultural land. The literatureuggests that landowners are more likely to give associations landhat is of poorer quality (Verdery, 2004). Therefore, one could arguehat in regions with higher land quality, the probability of landown-rs farming individually is higher.

The distance from the commune to the closest city is a proxyor access to local markets. The average distance in the communesncluded in the survey sample is 18 km. Nevertheless, there is quitesignificant variation across the 45 villages. Easier market access

owers the cost of landowners to commercialize their products andncreases access to information on market conditions (i.e. price andemand), which should facilitate individual farming.

The share of arable land in the commune aims to capture theoncentration of agriculture in the commune. More arable land

uggests that there is a higher potential for large-scale agricul-ure, which would make associations or leasing-out more attractivelternatives.

In addition to these variables I included a region dummy variableo control for regional differences, and commune fixed effects to

2e

26 (2009) 1080–1089 1085

ontrol for institutional differences that could not be accounted fortherwise.

esults

Table 5 reports the results of the probit regressions for the twoecision levels in 2006. The dependent variable in Model I is: yI = 1

f the landowner farms all the land individually, and yI = 0 if theandowners uses alternative farming arrangements (associationsnd/or leasing-out). The dependent variable in Model II is: yII = 1 ifhe landowner farms land in associations, and yII = 0 if she leases outhe land. Below I describe the results for the two decision levels.2

hoosing between farming the land individually and joininglternative farming arrangements

I find that for older households, the probability of farming theand individually is significantly higher as compared to choosinglternative farming arrangements. Nevertheless, age has a non-inear effect; as the age of the household head increases from 40 to0 years, the probability of farming individually starts decreasingor those that are older than 60–70 years. This finding is interest-ng in that it reinforces how important agriculture is as a source ofncome for the majority of the rural population. In addition it pointso the lack of alternative sources of revenue in the countryside,esulting in longer years of active farming.

A decade ago, in 1996, the relationship between age and therobability of farming the land individually was different (Vidican,008). Older households were more likely to seek alternative farm-

ng arrangements but after 70 years landowners chose to farmndividually. The non-liner effect of age is worthy to note becauset suggests that earlier in the transition we might have observedhe phenomena of “collective memory” as suggested by the path-ependent argument (Hann, 1993). Based on this argument, olderouseholds who experienced the coercive character of collectiviza-ion and lived thorough the inter-war period when property rightsere granted (even if for short periods of time in some regions)

nd the struggles for land were the defining features of rural life,ished to maintain full control of their property rights following

estitution (Vidican, 2008).The coefficients for years of completed education show that

ouseholds where the household head is more educated, are moreikely to seek alternative farming arrangements, suggesting that

ore education increases the opportunity for outside job opportu-ities, creating disincentives for farming all the land individually.

Households with a higher endowment of land per adult areore likely to engage in alternative farming arrangements, which

s not what I initially expected. As the land–person ratio increasesy one unit, the probability of engaging in alternative farmingrrangements increases by 0.04, significant at 5% level. Hence, onlyouseholds with surplus of land beyond what they need to secureminimum standard of living contract out land to associations or

ease to other private farmers. Since land is the main source ofncome and food for the majority of the rural population, land real-ocation occurs only for households with a higher endowment ofand relative to household size. These findings also suggest that

2 For this paper I choose to focus mainly on the results using the data from the006 household survey. Analyses of the 1996 household survey can be found in Rizovt al. (2001) and Vidican (2008).

1086 G. Vidican / Land Use Policy 26 (2009) 1080–1089

Table 5Estimation results—probit regression for the two decision levels (marginal effects).

Model I Model IIIndividual farming (1) vs. alternative arrangements (0) Associations (1) vs. leasing-out (0)

Household characteristicsAge .021** (.009) .037** (.015)Squared age −.0002** (.000) −.0003** (.000)Education −.014* (.007) .040** (.016)Land–adult ratio −.043** (.019) .042 (.034)Agriculture main occupation −.068 (.057) .088 (.077)

Farming conditionsLand fragmentation .054 (.040) .310*** (.082)Title −.089 (.103) .257** (.089)Household labor .001** (.000) .001** (.001)Loan dummy .141* (.066) .059 (.123)Tractor dummy .208** (.081) −.034 (.356)Seeds dummy .231** (.068) −.317*** (.088)Livestock .012 (.014) −.008 (.024)

Commune characteristicsLand quality −.004** (.002) .006 (.006)Distance to markets −.006 (.006) −.031 (.019)Share of arable land .003 (.006) −.007 (.011)Dummy-Western Plain .3763 (.216) −.351* (.166)Commune fixed effects Yes YesPseudo R-squared 0.453 0.437Observations 618 192

Note: clustered standard errors in parentheses. The table reports the coefficients and the standard errors with various independent variables, namely the age (and age squared)of the household head, the number of completed education years for the household head, the ratio of owned land to the number of adults in the family, dummy variable forwhether agriculture is the main occupation, the log of land fragmentation normalized by owned land, a dummy for whether the household has ownership title, the share ofhousehold labor, whether the household took loans, a dummy for whether the household owns a tractor, a dummy for whether the household purchased seeds, land qualityat commune level, distance to markets, share of arable land in the commune, and a dummy for the Western Plain region. Source: Calculations based on the 2006 households

perueF

maao5edf

FrS

at

mnsol

lt

urvey in the Western Plain (WP) and Central Romanian Plain (CRP) using Stata.* Significance level p < .10.

** Significance level p < .05.*** Significance level p < .01.

robability of choosing to farm the land individually at different lev-ls of land–adult ratio. The trend shows that at higher land–adultatios, households in the CRP are less likely to farm the land individ-ally as compared to households in the WP. Given the less favorableconomic conditions and demographics in the CRP (see Table 2),ig. 1 suggests that individual farming might be more costly.

Households where the household head has agriculture as theain occupation are more likely to seek alternative farming

rrangements (although not statistically significant), which runsgainst my initial hypothesis. However, the high age levels (mostf those with agriculture as a main occupation, 68%, are between

6 and 78 years old) and limited physical ownership contribute toxplaining these differences. Higher education, as showed above,oes also predict an increased likelihood of seeking alternative

arming arrangements. Nevertheless, 77% of those with agriculture

ig. 1. The probability of choosing to farm individually and land–adult ratio by agro-egion in 2006. Source: Calculations based on the 2006 household survey, usingtata.

itoce

asf

caupoh

malifi

s main occupation, have less than 8 years of education, suggestinghat age and capital availability explain better this result.

As Table 5 illustrates, households with a high degree of land frag-entation were more likely to farm the land individually, although

ot statistically significant. This suggests that, due to the sub-istence character of agricultural production, land fragmentationffers a risk spreading strategy for households wishing to farm theand themselves.

Access to physical capital is another critical factor that affectsand reallocation decisions. Therefore, it is not surprising to findhat households owning a tractor are more likely to farm the landndividually. The results show that in 2006 owning a tractor raisedhe probability of farming individually by 0.23. Similar results werebtained for households that purchased seeds and for those thatontracted bank loans, as indication of availability of capital andntrepreneurship.

Livestock plays a limited role in the decision process for landllocation between different farming arrangements. Although nottatistically significant, higher livestock increases the likelihood ofarming individually, against my initial hypothesis.

The coefficient for the land quality variable indicates that in theommunes with a higher share of good quality land, householdsre more likely to seek alternative farming arrangements. This isnlike what the literature argues, suggesting that lack of capitalrevents households from farming individually when economiesf scale provided by alternative farming arrangements can offerigher productivity and returns.

Although not statistically significant, a longer distance to

arkets increases the probability of seeking alternative farming

rrangements. The estimated coefficient for the share of arableand in total land in the commune suggests that more arable landncreases the likelihood of farming individually, although the coef-cient is low and not statistically significant.

olicy 26 (2009) 1080–1089 1087

taltftsaei

Cl

agotllohlntclh

jrt0ota

irolir

leas

vfpopFmfpat

fm

Ffu

vdnrioilhalc

atlstooccfcIcadatfetoc

pwt5c

G. Vidican / Land Use P

To summarize, human capital does play an important role inhe choice between farming the land individually and choosinglternative farming arrangements. As hypothesized, age has a non-inear effect—the older the household head, the more likely she iso choose alternative farming arrangements. A similar outcome isound for more years of schooling, since higher education increaseshe likelihood of finding jobs in the non-agricultural sector. A higherhare of land per adult also predicts a higher probability of seekinglternative farming arrangements, unlike what I initially hypoth-sized. Lastly, availability of capital (both human and physical)ncreases the likelihood of farming the land individually.

hoosing between joining farming associations and leasing-outand

One interesting distinction between households that joinedssociations and those that leased-out land comes from the demo-raphic profile of the landowners. If we take age as an indicatorf ability and experience in farming, we observe that its effect onhe choice for land reallocation is two-dimensional. First, seniorandowners are more likely to join associations, while much olderandowners are likely to lease-out land. Moreover, the probabilityf joining associations is higher for households where the house-old head is more than 50 years old. But, beyond 70 years old,

andowners are more likely to lease-out land (as indicated by theon-linear effect through the squared value of age). Hence, givenhat in leasing-out arrangements the landowner does not have toontribute labor, leasing-out is the last resort for households thatack the ability to farm and associations are better suited for house-olds that can be more actively involved in farming.

The variable years of education measure the access to off-farmob opportunities as well as managerial skills and abilities. Theesults from Model II show that an increase in the years of educa-ion by one unit increases the probability of joining associations by.04, significant at 5% level. This suggests that lack of employmentpportunities outside the agricultural sector prevents landownerso participate in land markets making farming associations a betterlternative.

The ratio of land to adult members in the households shows thatn 2006 land rich households were more likely to join associationsather than to lease-out land. The result suggests that leasing-ut land is not an attractive option even for household with moreand, primarily due to high transaction costs and limited availabil-ty of non-farming income sources, as also indicated by qualitativeesearch (Vidican, 2008).

Households with agriculture as primary occupation were moreikely to join associations. This can be viewed as a self-help strat-gy. Landowners who want to remain active farmers can get easierccess to inputs, physical capital for the planting and harvestingeason, as well as improved access to distribution networks.

One of the most interesting findings, which have not been pre-iously discussed in the literature, refers to the degree of landragmentation. I found that households with more fragmented landlots are more likely to join associations rather than to lease-ut land. A one-unit increase in land fragmentation increases therobability of joining associations by 0.31, significant at 1% level.ig. 2 shows that at low levels of fragmentation, households areore likely to lease-out land. However, for households with more

ragmented plots, especially more than three different plots, therobability of joining associations is much higher. Hence, one can

rgue that decisions on land reallocation reflect a response to mul-iple dimensions of fragmentation.

As I discussed in the previous section, when choosing betweenarming the land individually or in associations, households with

ore fragmented farms were more likely to farm the land indi-

sp

eh

ig. 2. The probability of joining associations and leasing at different levels of landragmentation in 2006. Source: Calculations based on the 2006 household survey,sing Stata.

idually. One could claim that the results for the two stages ofecision are contradictory. Nevertheless, there are two main expla-ations that show this is certainly not the case. First, the secondesult (choosing between joining associations or leasing-out land)s conditional on the first (choosing between farming individuallyr seeking alternative farming arrangements). Second, and moremportantly, given that membership in associations often requiresabor contribution (not only land), joining associations allows mostouseholds to farm the land themselves while the farming associ-tion assists them with physical capital (and sometimes seeds) forand preparation, seeding, and harvesting, and often times withommercialization.

This result brings to light an interesting difference betweenssociations and leasing-out arrangements. Due to high transac-ion costs for leasing-out (i.e. land registration and notary fees),and consolidation through the land market seems difficult at thistage. Land registration is not fully completed in most areas ofhe country (although significant progress has been made), andther factors related to social stratification and embeddednessf these institutional arrangements in the local economy makeonsolidation through leasing a more tedious process. Most asso-iations were created on the structures of the formal collectivearms, which facilitated farming on larger plots without additionalosts related to contractual arrangements and land registration.n addition, as Meurs (2001) argues, there is an additional socialost emerging from the unfamiliarity with new arrangementsnd market circumstances, which farmers are factoring in theirecision-making process. It is likely that as land markets developnd landowners become more accustomed with these transac-ions, consolidation through leasing becomes a closer alternativeor landowners. Therefore, policies that address the problem ofxtreme land fragmentation should look at farming associations ashe primary channel for consolidating dispersed land plots. More-ver, land registration should be completed in order to reduce theosts associated with land market participation.

Having title to the land in ownership does not predict partici-ation in the land market. Rather, households with ownership titleere more likely to join associations (having land title increases

he probability of joining associations by 0.26, significant at%). Having title for the land in ownership reduces transactionosts for households willing to lease-out land. Hence, this resultuggests that other factors might contribute to the decision of not

articipating in land markets.

A higher availability of household labor predicts that landown-rs are more likely to join associations (significant at 5% level). Also,ouseholds that contracted bank loans during the past year are

1 olicy

mi

calSt(b

rloi

stcptoh

atthatm

C

tccestoi

pckbalatpitispcfaasbj

ato

itbegisba

wfpnnsletts

A

fttrtr

R

A

A

B

B

B

C

D

D

D

D

D

D

E

088 G. Vidican / Land Use P

ore likely to join associations (although not statistically signif-cant).

The coefficients for the tractor and seeds dummies, as proxies forapital availability, show unexpected results. Households that owntractor and those that purchase seeds are more likely to lease-out

and, although statistically significant only for the seeds variable.ince capital availability is also an indication of entrepreneurship,his result suggests that households who invested in these itemstractor and seeds) have a more entrepreneurial profile, as reflectedy their participation in land markets.

The coefficient for the livestock variable, as a measure ofesource endowment, suggests that households that have moreivestock (as measured in standard heads) are more likely to lease-ut land (although not statistically significant), reflecting a changen occupational specialization.

As far as the commune characteristics are concerned, where thehare of good quality land is higher, households are more likelyo join associations. When the distance from the commune to thelosest city is higher, households are more likely to lease-out land,rimarily because of higher transportation costs incurred by gettinghe products from the farm to the market. Lastly, a higher sharef arable land in total agricultural land in the commune increasesouseholds’ likelihood of leasing-out land.

To summarize, the strongest predictors for household joiningssociations are age, education, and the degree of land fragmenta-ion. Leasing-out becomes a choice primarily for households wherehe household head is much older. I initially hypothesized thatouseholds with higher physical capital are more likely to joinssociations. However, the results showed otherwise, suggestinghat those with higher capital are more entrepreneurial, and hence,

ore likely to engage in market transactions.

onclusions

This paper examined why farming associations persistedhroughout the post-socialist transition period despite concerns forollective action problems and the availability of leasing-out as alose substitute. Land markets have been slow to develop despitexpectations that they will replace farming associations and willolve the land fragmentation problem in the Romanian agricul-ure. In addition, wide regional differences shaped the outcomesf land reform both in terms of economic performance, as well asnstitutional patterns for land reallocation.

The results from this analysis contribute to earlier studies ofost-socialist transition and de-collectivization by examining thehoice between joining associations and engaging in land mar-et transactions in addition to the more widely researched choiceetween farming individually or not. I find that the main factorsffecting land reallocation and the choice between associations andeasing-out land are: land fragmentation, age, and education. Inddition, qualitative research, discussed in Vidican (2008), supportshese findings and adds more nuances. Interviews with farmers andarticipant observations suggest that among the reasons for choos-

ng to join farming associations, the most important are: the abilityo retain full ownership title on the land; lack of resources for work-ng the land themselves (physical, financial, and human capital); aense of security and flexibility that the associations offer to com-ensate for market constraints (e.g. in-kind payment for mechani-al services); the older age of the rural population. In addition, otheractors such as the history of land reform and collectivization play

role in the post-socialist land reallocation decisions, discussedt length in Vidican (2008). Therefore, the results from this studyhow that in the decisions landowners make for reallocating landetween different farming arrangements (i.e. farming individually,

oining associations, leasing-out) the concepts of costs and benefits

F

H

26 (2009) 1080–1089

re much broader than what one may think. They refer not only tohe monetary component but also to aspects related to risk, securityf tenure, and historical considerations.

These combined findings suggest that despite wide critiquesn the literature and policy circles, and despite the decrease inheir number, formal associations offer some benefits to the mem-er households. The first such advantage is the ability to reachconomies of scale by consolidating land plots while maintainingreater control over land. Because economic diversification is lack-ng in the rural areas, few of the landowners are willing, or able, tohift out of agriculture. Moreover, farming associations appear toe a good option for landowners that are resource constrained butre still willing and able to be engaged in farming.

These findings have important policy implications at a timehen the agricultural sector in the post-socialist countries is con-

ronted with high inefficiencies. The myriad of small spread outlots prevent farm upgrading through investment and mecha-ization, within an overall economic and financial system that isot well adapted to small farming. Hence, farming associationshould be viewed as an alternative for achieving the much neededand consolidation, along with the opportunities created by thexpansion of land markets. Facilitating the access of associationso credit markets and supporting the channels for selling agricul-ural products are only some policies that could be considered fortrengthening their role for land reallocation.

cknowledgements

I am grateful to Tavneet Suri, Annette Kim, and Codrina Radaor their useful comments and suggestions on various drafts ofhis paper. I am also very grateful to an anonymous reviewer, forhe thorough and constructive suggestions and comments. Thisesearch was financially supported by the William Davidson Insti-ute Fellowship Grant (S-LMAQM-04-GR-0145). The author is solelyesponsible for any remaining errors.

eferences

llina-Pisano, J., 2007. The Post-Soviet Potemkin Village: Politics and Property Rightsin the Black Earth. Cambridge University Press, Cambridge.

ligica, P.D., Dabu, A., 2003. Land reform and agricultural reform policies in Roma-nia’s transition to the market economy. Eastern European Economies 41 (5),49–69.

entley, J., 1987. Economic and ecological approaches to land fragmentation: indefense of a much-aligned phenomenon. Annual Review of Anthropology 16,31–67.

larel, B., Hazell, P., Place, F., Quiggin, J., 1992. The economics of farm fragmentation:evidence from Ghana and Rwanda. World Bank Economic Review 6 (2), May.

rooks, K., Meurs, M., 1994. Romanian Land Reform: 1991–1993. Comparative Eco-nomic Studies 35 (2), 17–32.

arter, M.R., 1984. Resource allocation and use under collective rights and labormanagement in Peruvian coastal agriculture. The Economic Journal 94, 826–846.

eininger, K., 2003. Land markets in developing and transition economies: impactof liberalization and implications for future reform. American Journal of Agri-cultural Economics 85 (5), 1217–1222.

eininger, K., Songqing, J., 2003. Land sales and rental markets in transition: evidencefrom rural Vietnam. World Bank Policy Research Working Paper No. 3013.

iminescu, D., 1996. Deplasarile Osenilor in Strainatate: Un Nou Model de Migratie.Revista de Cercetari Sociale 2, 16–33.

obrescu, D., 1999. The Land Market and Other Related Issues Regarding Land Priva-tization in Romania. World Bank Consultant Report. The World Bank, Bucharest.

umitru, M., 2002. Country Case Studies on Integrating Land Issues into the BroaderDevelopment Agenda. Romania. The World Bank, Bucharest.

uncan, J., Prosterman, R.L., 2000. Land Market Development in Rural Romania. RDIReports on Foreign Aid and Development #107, Rural Development Institute,Seattle.

llickson, R.C., 1993. Property in land. The Yale Law Journal 102, 1315–1400.

oster, A., Rosenzweig, M., 1994. A test for moral hazard in the labor market: con-

tractual arrangements, effort, and health. Review of Economics and Statistics 76,213–227.

ann, C.M., 1993. From production to property: decollectivization and thefamily–land relationship in contemporary Hungary. Man, New Series 28 (2),299–320.

olicy

H

H

L

L

L

L

L

L

M

M

M

M

N

P

R

R

S

S

SS

S

S

S

S

V

V

V

V

tute of Technology. Doctoral Dissertation, Cambridge, MA.

G. Vidican / Land Use P

eiss, F., 2002. Structural choice analysis with nested logit models. The State Journal2 (3), 227–252.

ensher, D.A., 2005. Applied Choice Analysis: A Primer. Cambridge University Press,Cambridge.

erman, Z., 2005. Farm Fragmentation and Productivity. Evidence from Georgia.The Center for Agricultural Economic Research. The Department of AgriculturalEconomics and Management. Discussion Paper No. 8.05.

erman, Z., Cimpoies, D., 2006. Land consolidation as a factor for rural developmentin Moldova. Europe-Asia Studies 58 (3), 439–455.

erman, Z., Csaki, C., Moroz, V., 1998. Land Reform and Farm Restructuring inMoldova: Progress and Prospects, World Bank Discussion Paper 398.

erman, Z., Csaki, C., Feder, G., 2004. Agriculture in Transition: Land Policies andEvolving Farm Structures in Post-Soviet Countries. Lexington Books, Oxford.

erman, Z., Sedik, D., Pugachov, N., Goncharuk, A., 2007. Rethinking AgriculturalReform in Ukraine. In: Studies on the Agricultural and Food Sector in Centraland Eastern Europe, vol. 38. IAMO.

erman, Z., Shagaida, N., 2007. Land policies and agricultural land markets in Russia.Land Use Policy 24, 14–23.

ARD, 2007. National Strategic Plan for Agricultural Development. Romanian Min-istry of Agriculture and Rural Development (MARD), Bucharest.

anski, C., 1977. The structure of random utility models. Theory and Decision 8,229–254.

athijs, E., Swinnen, J.F.M., 2001. Production organization and efficiency during tran-sition: an empirical analysis of East German agriculture. Review of Economicsand Statistics 83 (1), 100–107.

eurs, M., 2001. The Evolution of Agrarian Institutions: A Comparative Study ofPost-Socialist Hungary and Bulgaria. University of Michigan Press, Ann Arbor.

IS, 2002. National Agricultural Census. National Institute of Statistics (NIS), Roma-nia, Bucharest.

ollak, R.A., 1985. A transaction cost approach to families and households. Journalof Economic Literature XXIII (June), 581–608.

izov, M., 2003. Human capital and the agrarian structure in transition: microevidence from Romania. In: 25th International Conference of Agricultural Eco-nomics, South Africa, Durban.

izov, M., Gavrilescu, D., Gow, H., Mathijs, E., Swinnen, J.F.M., 2001. Transition andenterprise restructuring: the development of individual farming in Romania.World Development 29 (7), 1257–1274.

V

W

26 (2009) 1080–1089 1089

abates-Wheeler, R., 2005. Cooperation in the Romanian countryside. An insight intopost-Soviet agriculture. Lexington Books, Colorado.

ahn, D.E., Alderman, H., 1988. The effects of human capital on wages, and thedeterminants of labor supply in a developing country. Journal of DevelopmentEconomics 29 (2), 157–183.

andu, D., 1999. Spatiul Social al Tranzitiei. Polirom, Iasi.cott, J., 1976. The Moral Economy of the Peasant: Rebelion and Subsistence in

Southeast Asia. Yale University Press, New Haven and London.langen, L.H., van Kooten, G., Suchanek, P., 2004. Institutions, social capital and

agricultural change in Central and Eastern Europe. Journal of Rural Studies 20,245–256.

tark, D., 1996. Recombinant property in East European capitalism. American Journalof Sociology 101, 993–1027.

uri, T., 2007. Selection and Comparative Advantage in Technology Adoption. MITSloan School of Management, Cambridge, MA.

winnen, J.F.M., Vranken, F., 2007. Patterns of Land Market Development in Transi-tion. LICOS Discussion Papers, Number 179/2007. Leuven, Belgium, LICOS Centrefor Institutions and Economic Performance, Katholieke Universiteit Leuven.

erdery, K., 1999. Fuzzy property rights, power and identity in Transylvania’sdecollectivization. Uncertain transition. In: Buroway, M., Vedery, M. (Eds.),Ethnographies of Change in the Postsocialist World. Rowman and Littlefield,Lanham.

erdery, K., 2003. The Vanishing Hectare: Property and Value in Postsocialist Tran-sylvania. Cornell University Press, Ithaca.

erdery, K., 2004. The obligations of ownership: restoring rights to land in post-socialist Transylvania. Property in question. In: Vedery, K., Humphrey, C. (Eds.),Value Transformation in the Global Economy. Oxford, Berg, pp. 139–159.

idican, G., 2008. Institutional Arrangements and Land Reallocation During Tran-sition: A Regional Analysis of Small Farms in Romania. Department of UrbanStudies and Planning, International Development Group. Massachusetts Insti-

ranken, L., Swinnen, J.F.M., 2006. Land rental markets in transition: theory andevidence from Hungary. World Development 34 (3), 481–500.

orld Bank, 2007. Romania—Farm Restructuring Project. Project Information Docu-ment. Appraisal Stage. The World Bank, Washington. URL: http://go.worldbank.org/XZK4O6JVU0.