How does practice matches land laws in Central Asia?

Akhmadiyeva, Zarema; Herzfeld, Thomas Article — Published Version How does practice matches land laws in Central Asia? Land Use Policy Provided in Cooperation with: Leibniz Institute of Agricultural Development in Transition Economies (IAMO), Halle (Saale) Suggested Citation: Akhmadiyeva, Zarema; Herzfeld, Thomas (2021) : How does practice matches land laws in Central Asia?, Land Use Policy, ISSN 0264-8377, Elsevier, Amsterdam, Vol. 109, https://doi.org/10.1016/j.landusepol.2021.105726 This Version is available at: https://hdl.handle.net/10419/242477 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ Land Use Policy 109 (2021) 105726 Available online 6 September 2021 0264-8377/© 2021 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). How does practice matches land laws in Central Asia? Zarema Akhmadiyeva a , * , Thomas Herzfeld a , b a Leibniz Institute of Agricultural Development in Transition Economies (IAMO), Theodor-Lieser-Straße 2, 06120 Halle (Saale), Germany b Martin-Luther-Universit¨ at Halle-Wittenberg, 06099 Halle (Saale), Germany ARTICLE INFO Keywords: Land rights Land intensification Tenure security ABSTRACT This paper aims to explore the association between discrepancies in land rights and actual practices on the one hand and farmers’ intentions to increase agricultural production on the other hand. We use farm-level data collected during a survey conducted in 2019 in southern Kazakhstan and eastern Uzbekistan and compare perceptions with land legislations of both countries. Comparative analysis of tenure conditions revealed that Kazakh farmers are less restricted in land use than Uzbek farmers. Moreover, insufficient law enforcement allows Kazakh farmers to violate existing restrictions on land transferability. Uzbek farmers tend to underuse their rights in such crucial issues as income generating and withdrawal of products from the land; the reason lies in inconsistency between land code and supplementary decrees for strategic crops. Our empirical findings show that mismatches between land rights on paper and perceived land rights −that might be the violation of law restrictions or incomplete use of land rights −generally reduces farmers’ willingness to increase production. 1. Introduction Land intensification that basically refers to the increased use of nonland inputs on a given plot plays an important role in improving global food security (Brookfield, 2001; Gustavsson et al., 2011). It comprises various farmers’ actions (often investment consuming) such as the conversion of fallow land to permanent cropland, an increase in inputs, and adoption of improved technologies. Land intensification like any other farmers’ decision is a complex process and requires preliminary actions such as problem detection, problem definition, analysis, development of intention, and implementation (Kay and Edwards, 1994; ¨ Ohlm´ er et al., 1998). Among these actions, individual intention is often used as a proxy for behaviour, although the implementation of this behaviour is not guaranteed. The complexity of the decision-making process and differences in institutional characteristics between developed and developing countries call for considering the role of institutional environments in farmers’ decisions. Several studies have found that the degree of tenure security is one of the most important determinants in farmer’s decisions on agricultural production and investment (Boserup, 1965; Holden and Otsuka, 2014; Lawry et al., 2014). However, contradicting evidence presented in empirical studies suggests that the effect of tenure security is not clear and dependent on the institutional environment (Brasselle et al., 2002; Newman et al., 2015). Besides, the lack of a common view on how to measure tenure security hampers the generalization of the tenure security effect. One of the pioneers who operationalize land rights as a security proxy was Timothy Besley (1995). He examined self-reported (de facto) transfer rights and found that improved rights facilitate land investment in Ghana. Later, Brasselle et al. (2002) extended the number of rights and included land use rights in their study. Twerefou et al. (2011) employed farmers’ de facto land transactions (land selling, leasing, and transfer) to measure their effects on land investment. Land titles associated with the formalization of land rights often fail to provide evidence of tenure security (Ma et al., 2015a). Ayamga et al. (2016) propose that security may vary with land tenancy arrangements. Authors found that farmers without ownership rights are likely to invest less in land improvement. Much of the literature on land tenure security uses formal or informal institutional arrangements separately to analyse their impact on the performance of farming activity, but neglects the effect of their interactions. This gap leads us to investigate how the interaction between legal land rights and perceived land rights affects land intensification. Introducing a new approach to compare what is in “paper and practice”, Klümper et al. (2018) assume that these paper-practice mismatches can lead to less investment in resource use; but authors do not provide corresponding empirical evidence. In countries where a large share of agricultural land belongs to the government, the main problem is that land rights are either not clearly defined or land users lack a consistent understanding of them. This * Corresponding author. E-mail addresses: [email protected] (Z. Akhmadiyeva), [email protected] (T. Herzfeld). Contents lists available at ScienceDirect Land Use Policy journal homepage: www.elsevier.com/locate/landusepol https://doi.org/10.1016/j.landusepol.2021.105726 Received 29 November 2020; Received in revised form 25 August 2021; Accepted 26 August 2021 Land Use Policy 109 (2021) 105726 2 uncertainty impedes the formation of land markets and distorts production incentives, investments in land-improving measures, and effective allocation of land resources (Feder and Feeny, 1991; World Bank, 2007). That is what has happened to Kazakhstan and Uzbekistan, two transition economies that individualized agriculture after the dissolution of the Soviet Union but failed to provide secure land rights due to ambiguous formulations of land law and ineffective law enforcement (Melnikovov´ a and Havrland, 2016; Hanson, 2017). In this study, we explore whether discrepancies between land rights and actual practices are related to farmers’ intention to increase agricultural production in Kazakhstan and Uzbekistan. The main contribution of our findings to the existing literature is that we use discrepancies between land rights written in national land codes and actual farming practices in empirical analysis for the first time. The remainder of this paper has the following structure: In the next section, we describe agricultural development and land tenure arrangements in Kazakhstan and Uzbekistan. Afterwards, we provide the conceptual framework of land rights and actual practices and their relevance to land intensification. Section 4 describes the methodology and data. Section 5 reports and discusses descriptive results on discrepancies between land rights and actual practices and empirical results of regression analysis. Section 6 summarizes the findings and draws policy implications for land administration in the focus countries. 2. Agricultural development and land tenure in Kazakhstan and Uzbekistan Kazakhstan and Uzbekistan are two neighbouring countries located in Central Asia. Both have a mix of arid and semi-arid climate and more than half of their territories are used under agriculture. In Uzbekistan, crop cultivation is mainly irrigated which is determined by environmental conditions such as high soil salinity and a long and dry summer. Due to the vast territory, Kazakhstan’s regions vary in environmental conditions. The northern part of Kazakhstan, where soil is relatively fertile, has rain-fed agriculture and the southern part is dominated by irrigated agriculture. While Uzbek people engaged traditionally mainly in sedentary farming, pastoral and nomadic Kazakh people were forced to sedentary farming during the Soviet era. However, the big part of agricultural land in Kazakhstan is still under pastures. The southern part of Kazakhstan together with Uzbekistan became a central component of cotton-growing region during the Soviet period (Petrick and Pomfret, 2016) and northern Kazakhstan became an important producer of wheat and barley. The production of predominant crops coming from the Soviet legacy remained after the collapse of the Soviet Union. To eliminate inefficiencies of collective land management after the dissolution of the USSR, Kazakhstan and Uzbekistan have undergone salient structural changes in the agricultural sector, albeit with significant differences. Recuperating from the initial slump caused by the disintegration of the centrally planned system, the two countries implemented a set of reforms to build their own national agricultural sectors. Both of them followed a so-called conventional approach to land reforms implying a redistribution of land used by state and collective enterprises to households and commercial farms (Lerman and Sedik, 2018). While Kazakhstan transferred, at least to some farms, ownership titles, Uzbekistan conserved state ownership of land (Melnikovov´ a and Havrland, 2016). Dehkans and farmers 1 received only land use rights with long-term lease contracts. 2 Moreover, the Uzbek government preserved the Soviet system of state quotas and subsidies for cotton and wheat production, denying farmers the opportunity to adjust land use to the new market environment. Kazakhstan went a bit further, undertaking the necessary measures to decentralize production and marketing processes and reducing the government’s influence to a minimum. Land ownership rights were granted to households in 1991 and commercial farmers in 2003. 3 Kazakh farmers have been exposed to fewer state interventions and faced more market-driven incentives. Despite of political and economic reforms intending to establish a market economy, substantial structural changes (such as outmigration of rural population, disruptions in supply chains, removal of agricultural subsidies) resulted in a substantial decline in land use and agricultural production. According to the numbers reported by the Food and Agriculture Organization of the United Nations, about 20 million hectares of cropland in Kazakhstan were abandoned from 1991 to 1999 and only 6 million hectares were recultivated by 2018. Uzbekistan, in contrast, succeeded to avert a dramatic collapse in agricultural production and removed only 800 thousand hectares of sown area in the first decade of independence, of which 300 thousand were recultivated again in the next two decades. Land reallocation resulted in an increase in the share of individual farms in agricultural production (Lerman and Sedik, 2018). Between 1991 and 2010, the contribution of individual farms to Gross Agricultural Output raised from 32 to 71 per cent in Kazakhstan and from 33 to 98 per cent in Uzbekistan. The cropland productivity of newly established individual farms was not significantly different from cropland productivity of agricultural enterprises at the beginning of 1990 s (Fig. 1); yet, after 2005, individual producers outperformed corporate farms and continued to enlarge the disparity. Despite the considerable contribution of individual farms to the agricultural sector, they still faced constraints inhibiting their incentives. The land reform process aiming at better tenure arrangements remained incomplete in Kazakhstan. Inadequate institutional capacity, high transaction costs of land exchange, skewness of land distribution towards large-scale producers left land market inactive (Kvartiuk and Petrick, 2021). As a result, Kazakh farmers still prefer to rent state land paying low fees instead of acquiring private ownership. 4 Moreover, prioritized crops were still produced by enterprises that dominated in sub-sectors. In February 2013, Kazakhstan adopted a new programme 1 In Uzbekistan, the term dehkan farm designates a small-scale household farm and farmer is a land user with larger farm size relying on hired labour. 2 According to the Land Law of Uzbekistan (1998), the duration of lease contract can be up to 50 years, but not less than 30 years; “leased land parcels cannot be objects of sell - buy, mortgage, present, exchange. The right to lease the land parcel can be passed as a mortgage for getting credits. The lessee has the right to pass his lease rights to the land parcel as a mortgage without the consent of the lease-giver only in cases envisaged by Law or the lease agreement.” However, with amendments to the Land Code introduced on 29.09.2020, commercial farmers may sublease their plots up to one year for agricultural purposes. The lessee has a primary right to renew the land lease contract after its expiry. 3 In Kazakhstan, households are small-scale land users that mostly engage in labour-intensive and livestock production to meet their own consumption needs on land located in rural and suburban areas. Commercial farms are agricultural producers that operate in the form of individual, family, or corporate farms. Generally, these farms are larger and market most of their production. 4 According to Kazakh Land Code (2003), farmers may lease land for up to 49 years without the rights to sell or lease. However, they have a priority to renew a lease contract or purchase the land. The mortgage of agricultural land is illegal for leaseholders. Z. Akhmadiyeva and T. Herzfeld Land Use Policy 109 (2021) 105726 3 “Agribusiness 2020”, the main objectives of it were boosting agricultural production and enhancing the competitiveness of local producers through specific measures, such as improving the effectiveness of state regulation, the financial rehabilitation of agricultural sectors, and the removal of bureaucratic barriers (Agribusiness-2020, 2012). However, international and local experts have expressed their concerns about the effectiveness of government support, because farmers needed not only eased access to inputs and capital but also the freedom to discover productive market opportunities and trust to institutions. Therefore, they highlighted the need to give more attention to the institutional and regulatory framework that would enable market mechanisms to allocate resources effectively (Petrick and Pomfret, 2016; Oshakbayev et al., 2018). Uzbekistan has undertaken several agricultural reforms aiming at dismantling large-scale farms: farm reorganization in 1992 and farm fragmentation in 1998 (Djanibekov et al., 2012). The farm restructuring, however, neglected the fact that the irrigation water supply system was designed to serve large scale farms during the Soviet period. Excessive fragmentation of fields resulted in delays of water supply that, consequently, led to disruptions in agricultural production. The state initiated a new wave of farm consolidation in 2008 to readjust farm sizes that would be suitable for the existing irrigation system. Farmers with less than 30 ha returned their lands on an involuntary basis without receiving any compensation (Melnikovov´ a and Havrland, 2016). Cotton and wheat production that occupies more than 80 per cent of the sown area and remains under the state order system is subject to another consolidation reform since 2018 (Zorya et al., 2019). All of these reforms seek to enhance productivity, but such frequent farm reorganizations have undermined farmers’ management and investment incentives increasing land tenure insecurity among farmers (Mukhamedova and Pomfret, 2019). Implementation of land reforms in Kazakhstan and Uzbekistan is inextricably linked to the level of law enforcement. If land legislation is not sufficiently enforced and does not involve adequate monitoring and punishment for law violations, the assurance of rights is low (Sjaastad and Bromley, 2000). Poor detection of violations and the low severity of punishment increase the incentive to break rules that, in turn, leads to increase in tenure insecurity. Kazakhstan and Uzbekistan impose different levels of law enforcement on land users. Given that agriculture is the most centralized sector in Uzbekistan, the government strictly controls the majority of agricultural activities including input and output allocation, and enforces all relevant state decisions. Particularly, cotton and wheat producers risk to be subject to administrative and criminal liability and to lose their land if they fail to fulfil mandatory targets (Muradov and Ilkhamov, 2014). In Kazakhstan, where agriculture has the lowest share in the Gross Domestic Product, the overall economic system is closer to a market economy than in Uzbekistan, but the ability of state structures to enforce effective economic policies remains weak (Satpayev, 2014; World Bank, 2018). Due to the inconsistency between central and local governments, the implementation of approved state programs often does not start or remains incomplete at the regional level. Local government bodies can disregard national legislation and state orders in the pursuit of self-interest or the interest of local elites. The lack of law enforcement, hence, is a result of perpetual competition between different government levels for the redistribution of administrative power, and ineffective mechanism for inspection and criminal prosecution. Such low quality of institutions and incomplete land markets push farmers to respond with risk-reducing activities. Therefore, we assume that law enforcement is strongly associated with the perception of land rights by farmers. 3. Conceptual framework 3.1. Bundles of land rights and actual practices Providing tenure security is important for creating incentives for investing in land. When governments fail to establish secure tenure conditions, farmers tend to sacrifice future returns to protect themselves against adverse income shocks by using inputs less intensively (Morduch, 1995). Empirical studies have confirmed the positive association between tenure security and production decisions (Feder et al., 1988; Besley, 1995; Ma et al., 2017); however, there are still debates on what is mentioned by the term ‘tenure security’. Sjaastad and Bromley (2000) distinguished two concepts regarding land tenure: substance of rights referring to the content of land rights and assurance of rights referring to uncertainty of rights. The content of the bundle of rights is included in a land title and defines the privileges to use a given plot to one user and assign duties on others to stay off this land. Moreover, access to credits and the activity of land markets depends on the substance of land rights. Scott (2008) identifies “three powers” given by land rights: the power to use and manage land, the power to transfer and alienate it, and the power to take income or rent from land use. In a similar but more detailed approach, Schlager and Ostrom (1992) propose to split property rights into two groups: operational-level rights, which enable Fig. 1. Productivity of individual farms and agricultural enterprises (1995–2014). Compiled by authors using data from statistical yearbooks and Penn World Table (9.0). Z. Akhmadiyeva and T. Herzfeld Land Use Policy 109 (2021) 105726 4 individuals to access land and withdraw the land’s products, and collective-choice level rights, which grant authority to elaborate operational-level rights. They developed a conceptual scheme to distinguish five independent bundles of rights (access, withdrawal, management, exclusion, and alienation). Pooling together these classifications, Klümper et al. (2018) use three main bundles of land rights to analyse households’ tenure conditions in Tajikistan; each of the main bundles contains case-dependent rights that can be combined under a common characteristic: 1) Land use rights include access, withdrawal, and land use change rights. 2) Control and decision-making rights include management, investment, exclusion, and income-generating rights. 3) Alienation rights include reallocation, sell, leasing, and inheritance rights. Some scholars claim that the right to tenure security and government protection is attached to land title and granted by the authority issuing official land titles (Place, 2009; Ma et al., 2017). Others see the security as a separate full-fledged right in the joint bundle of tenure rights. In the seminal work of Honore (1961), the right to security has a place in the list of the eleven “incidents of ownership”. Following several studies that attempt to revisit and improve the Schlager-Ostrom framework (Galik and Jagger, 2015; Sikor et al., 2017; Klümper et al., 2018), we add the bundle of government protection that consists of the protection of tenure rights by courts and the power of land certificates. The right to protection by courts allows farmers to protect own interests in courts in disputes with other farmers, foreign investors, or local authorities. The power of land certificates helps farmers to confirm land rights documentally. Land certificates should give security to the holder by default, but the farmer’s perception of certificate validity can differ completely from what is originally conceived. This may be a consequence of the insufficient implementation of land regulations at the local or regional level, or of the farmers’ mistrust in the administration and government to uphold and protect their rights. Table 1 presents the four bundles of rights used in this study. Sjaastad and Bromley (2000) insist that assurance of rights is sufficient for the concept of tenure security that, in turn, can be defined as the perception of the likelihood of losing a specific land right. Ma et al. (2017) reckon that it is perceptions of land rights, not land certification or laws, that impact directly on land users’ decisions. 5 Perceptions affect decisions about land use and result in the actual implementation of land rights that can be inconsistent with the law (Deininger, 2003). Therefore, land use patterns are determined not only by legal rights (LR) but also by actual practices (Ellickson, 1986; Klümper et al., 2018). Actual land use practices are shaped not only by perceptions of land rights but also by perceptions of informal institutions such as customary rules or commonly accepted practices (Prell et al., 2010). Commonly accepted practices can be a result of either traditional farming practices or the inability of farmers to adapt to socio-economic and political transformations, such as transition from plan to market economy or technology shocks, and to adjust their farming practices creating new combinations of formal and informal institutions (Ward and Kingdon, 1995; Chavance, 2008). For instance, in some former Soviet countries that had no recognition of private land ownership during the collectivization period, farmers faced challenges after decollectivisation due to the absence of the skills and farming practices necessary for private farming (Rozelle and Swinnen, 2009). Therefore, they might not be eager to acquire land ownership. In the context of this study, we use the term ‘actual practices’ (AP) to cover informal rules and informal farming practices (e.g., illegal subleasing or land reallocation) emerging with decollectivisation and land reforms. 6 3.2. Land use and discrepancies between land rights and actual practices Legal rights and actual practices may support (LR=AP) or contradict each other (LR∕=AP) (Klümper et al., 2018). Farming practices supported by the relevant property right bear fewer risks for farmers and create a secure environment for land use. Contradicting LR-AP combinations could be two-folded. The first contradiction appears in situations when legal rights are more pronounced than actual practices (LR>AP) and might be a result of farmer’s unawareness about land rights or low exploitation of these rights by farmers. The second contradiction appears in situations when actual practices exceed the relevant rights (AP>LR), which implies that farmers violate legal restrictions, ignoring the authority of local administration and the rule of law. That might be the case when non-compliance is not sanctioned by effective governance mechanisms and actual land tenure practices are likely to fill gaps of the legal system. This type of mismatches also indicates weak institutions with insufficient law enforcement and a lack of trust in government (Broegaard, 2005). The relationship between LR-AP discrepancies and land users’ decisions has not been examined empirically in extant literature. Following the claims by Klümper et al. (2018) about the relationship between non-compliant behaviour and insecurity, three scenarios can be identified which would imply different incentives for a farmer’s intention to intensify land use: Table 1 Bundles of land rights and descriptions. Bundles Rights Description Land Use Access Right to enter a defined physical plot Withdrawal Right to obtain the benefits from land Land use change Right to change the type of agricultural activity Control and decision-making Management Right to control internal use patterns and transform the land by making improvements Investment Right to invest in land melioration and irrigation systems Exclusion Right to define who has access to the land Income generating Right to earn income from the land Alienation Reallocation Right to sell or lease the right of management and/or the right of exclusion Sell Right to sell the land Leasing Right to rent out the land Inheritance Right to inherit the land Government protection Protection by courts Right to government protection by courts Power of land certificates Right to have legally valid land certificates Note: definitions were compiled from Schlager and Ostrom (1992), Meinzen-- Dick (2014), Klümper et al. (2018). 5 Perception of tenure security has been increasingly used in addition to substance-based measurements not only in regional empirical studies. Considering the importance of perceived tenure security, international organizations also gather corresponding indicators to monitor true tenure security across countries. For example, the Sustainable Development Goals indicator framework adopted by United Nations developed Indicator 1.4.2 that incorporates perception of “tenure rights as legally secure, regardless of whether these rights are documented.” Another example is the Prindex program, initiated by Global Land Alliance and the Overseas Development Institute that launched a global survey on perceived tenure security. 6 We avoid to use the term “customary” rules that is often mentioned in studies that focus on the African countries (Mwangi and Meinzen-Dick, 2009; Meinzen-Dick, 2014; Holden and Ghebru, 2016). The reason is that it is not relevant in the context of Central Asia because crop cultivation does not arise from the community’s traditions over a long period or the use of land by ancestral societies. Z. Akhmadiyeva and T. Herzfeld Land Use Policy 109 (2021) 105726 5 1) Actual practices are fully aligned with the formal legal framework (LR=AP). We use this as default scenario. 2) Actual practices might not fully exploit the opportunities offered by the legal framework (LR>AP). Although we cannot conclude whether a farmer doesn’t use opportunities out of free decision or due to conflicting agricultural policy instruments such as the Uzbek procurement policy, we hypothesize that this farmer’s intention to intensify will be lower as in the default scenario. 3) Actual practices might exceed the legally provided opportunities (AP>LR). By definition, formal land laws must imply certain restrictions of a respective right in order to result in such a discrepancy. Conditional upon a farmer’s awareness of this non-compliant behavior, her/ his intention to intensify land use might be higher (unaware of legal rights) or lower (aware on non-compliance) than in the default scenario. In the first case, a farmer might expect to possess more rights than foreseen in the land code and might assume to reap the benefits of an intensified land use. In the latter case, the farmer is assumed to be aware of the risk of punishment, e.g. losing rights or future benefits from land. Therefore, farmers’ incentives to intensify land use is expected to be less than in the default scenario. However, two reasons prevent the development of hypotheses involving a causal relationship between discrepancies and the intention to intensify land use. First, as some studies provide evidence of high perceived tenure security even in the absence of formally established property rights (UNHR, 2015; Rao et al., 2017), the extent of law enforcement represents an important contextual variable. Second, both perceptions, of actual practices and intensification, are not formed independently from each other. Thus, any assumption of a causal relationship will be impossible to be falsified econometrically. This study provides only additional insights on how far farmers’ perceptions about land rights deviate within groups with different tenure settings. Therefore, the further analysis carries only an explorative character. 4. Methodology and data 4.1. Study area and sample selection We use data from a farm survey 7 conducted in March and April 2019 in two different neighbouring regions with irrigated agriculture: Turkistan province in southern Kazakhstan, and Samarkand province in eastern Uzbekistan. Three districts were chosen from each province to select respondents. In Uzbekistan, 460 farms −that constitute 30 per cent of the farm population −were selected randomly from 1525 eligible farms 8 in Pastdargom, Payarik, and Jomboy districts. A two-step sampling approach was applied in Kazakhstan. As Kazakhstan’s three districts −Maktaaral, Shardara, and Sariagash −have 31 sub-districts, three sub-districts were chosen randomly in each district. Further, in each sub-district, around 50 farms were selected randomly resulting in 495 farms in total, representing 2 per cent of all farms in three districts. Fig. 1A in the Appendix demonstrates aggregated land area used by farmers interviewed in the survey. The land is grouped by tenure conditions in two study regions. Uzbek respondents rent state land that constitutes 17914 ha or 4 per cent of the total sown area in Samarkand province. The area farmed by respondents in Kazakhstan sums up to 6485 ha which constitutes 0.8 per cent of the total sown area in Turkistan province. Of 6485 ha, 55 per cent is under private ownership, 30 per cent is rented state land, and 15 per cent is rented land from other farmers. 4.2. Description of variables of interest 4.2.1. Intention to increase production Intention and its relation to behaviour have been investigated by various studies. The Theory of Reasoned Actions and its extension, the Theory of Planned Behaviour (TPB), propose a useful framework to understand farmers’ behaviour based on their perceptions and beliefs (Ajzen and Fishbein, 1980; Ajzen, 1991). These theories suggest that the most proximal predictor of behaviour is intention. Lefebvre et al. (2014) investigated farmers’ intention to invest in land and found that the majority of farmers (74 per cent) demonstrate behaviours consistent with their intentions. According to Viira et al. (2014), farmers’ intentions predict effectively actual behaviour when the considered action is positive (farm growth) rather than negative (farm shrinkage). Given these aspects and that the value of agricultural intensification is not observed directly in this study, we use farmers’ intentions to increase production as a proxy for land intensification. To collect data on intentions, farmers responded to the question: “How likely is it that you will increase crop yield in at least part of your farm in the next year?” The variable is ordinal and measured on a five-point Likert scale anchored with 1 =Extremely Unlikely and 5 =Extremely Likely. Fig. 2A in the Appendix demonstrates that the patterns of responses are similar in both regions. The distributions of responses are skewed to the left, indicating that most of the farmers have positive attitudes towards increasing production. The willingness to gain higher output can be characterized as a socially acceptable option; therefore, one might doubt the validity of responses due to social desirability bias (Nederhof, 1985). To reduce the bias as much as possible, interviewers notified farmers that there was no right or wrong answer, and the data collected would be treated confidentially and in aggregated form. The Likert scales were placed at the beginning of the questionnaire to reduce bias from participant fatigue. All respondents and interviewers were men that excludes the risk of gender-related bias. We examine the convergent validity of this variable with two additional variables that also measure the willingness to intensify land use on the five-point Likert scale. 9 These questions were formulated following the procedures for constructing the TPB questionnaire recommended by Fishbein and Ajzen (2010). Using multiple items to measure a single concept allows covering different aspects of the concept and reducing the measurement error that can originate from poor wording or misunderstanding of the scaling approach (Hair et al., 2017). The average variance extracted, a criterion of convergent validity that refers to the degree to which measures are related, is 0.544 that is above the conventional threshold of 0.5. The Cronbach’s α of three variables is 0.78 (recommended threshold is 0.7) and indicates a high internal consistency. 4.2.2. Operationalization of land rights, actual practices and their discrepancies We use a methodological approach applied by Klümper et al. (2018) to operationalize land rights and actual practices in order to calculate discrepancies between them. Table 2 presents four bundles of rights as well as the coding of legal rights and responses to questions covering actual practices. The results of pre-survey explorative interviews revealed that farmers do not intend to rent their land out due to land scarcity, but instead, many would like to lease more land. Therefore, we split the lease right into two, the right to rent out and the right to rent from farmers who lease state land (land tenants), to mirror the actions of 7 The Agrichange II survey has been financed by Volkswagen Foundation, German Ministry of Education and Research (BMBF), and IAMO. 8 Eligible farms in this study are legally registered farms that produce mainly crops, have at least 80% of irrigated land, and were active in farming in 2019. Respondents were chosen from the list at regular intervals. 9 For two additional variables, we used the following statements: “Do you plan to increase crop yield in at least part of your farm in the next year?” and “How strong is your intention to increase crop yield in at least part of your farm in the next year?” Z. Akhmadiyeva and T. Herzfeld Land Use Policy 109 (2021) 105726 6 the latter ones and to identify if they violate legal restrictions. The operationalization implies converting the land rights and the perception of these rights (actual practices) by farmers into measurable factors. To analyse actual practices, we use survey data that encompasses farmers’ self-assessment of their land rights. A 5-point ascending scale (see Table 2 column 4) was offered after each question formulated as: “To what extent are you free to use the right to access, withdraw from land, etc.?”. Legal land rights were assessed on the basis of the Kazakh and Uzbek national land codes. To assign a score to land rights, we used the same 5point scale but points 2 and 4 were intentionally omitted to avoid difficulties in assessing legal restrictions in land use. Instead of classifying limitations into three levels (low, medium, and high), we introduced one medium level. Score 1 was assigned in case if a right is completely prohibited by law, score 3 was given to rights with partial restrictions, and score 5 was assigned to full rights. We identified three categories of farmers who participated in the survey and have different legal privileges and limitations in land use: (1) Uzbek land tenants, (2) Kazakh landowners, and (3) Kazakh land tenants. The assessment of their land rights and the respective articles in the national land codes are presented in Table 3. As the results show Kazakh farmers with private farmland (left bar chart) should have the highest amount of fully transferred land rights among three groups as indicated by a score of 12 out of 14. The two exceptions are the limited right to land use change and the right to lease from farmers with rented state land. Kazakh farmers who rent state land (middle column) face more legal restrictions than those who own land. Particularly, all rights in the alienation bundle, except for the inheritance right, are completely restricted. Finally, Uzbek farmers (right column), the most deprived group in terms of legal land rights, do not have any right in the bundle of alienation rights and the right to land use change. Besides, their management right and the right to invest in land improvements are limited. Before undertaking any change in land management or investing in land improvements, Uzbek farmers are obliged to obtain permission from the local authorities. Later on, these changes have to be carried out under the control of respective institutions. The presence of written land law and official land titles implies that the legitimacy of these documents is fully backed by authorities who issued them. Therefore, two rights in the government protection bundle (the right to government protection in courts and the certificates’ importance) are assigned the score 5 for all three groups. After the quantitative assessment of land rights and actual practices, we calculate their discrepancies using the following equation: Discrepancy =Land Right −Actual Practice (1) The scale of discrete discrepancy values may range from -4 to 4, where negative values certify a potential law violation, i.e. rights overuse. Positive values show the underuse of right and appear when the farmer responds not to have the right to the allowed extent; in other words, the farmer feels restricted to perform an activity regulated by law (due to insufficient knowledge, controversial regulations, arbitrariness of local officials, etc.). The absence of discrepancy, 0, means that land rights perfectly overlap with actual practices. Table 4 provides details on the discrepancies between land rights and actual practices for Kazakh and Uzbek farmers. Overall, 21 different discrepancies were retrieved from comparing the legal assessment and survey respondents’ statements. As can be seen from the table, the list of discrepancies differs for Kazakh and Uzbek farmers, in some cases even to a large extent. For ease of understanding, descriptive statistics of positive and negative discrepancies are displayed separately; the mean and standard devastation of discrepancies are calculated for different subsamples the size of which depends on the number of farmers with negative or positive discrepancies. Considering that the range of each discrepancy is four at maximum, the standard deviation in discrepancies in the alienation bundle is quite high for Kazakh farmers, more than one. Uzbek farmers show high variation in discrepancies for the rights to withdrawal, exclusion, and income generating. 4.2.3. Additional explanatory variables Control variables included in the empirical analysis comprise farm and farmer characteristics. Farmer-specific variables incorporate educational level, age, special agricultural education, use of consultancy services, and the desired period of land use in the future. These factors Table 2 The operationalization of land rights and land-use practices. Bundles Right/Practice Legal rights Actual practices Land use Access Withdrawal Land use change 1-no right; 1-Never hold the practice; Control and decisions Management Investment Exclusion 2-rarely hold; Income generating 3-limited right; Alienation Reallocation 3-occasionally; Sell Renting out 4-very frequently; Leasing from land tenants Inheritance 5-full right. 5-always. Government protection Protection by courts Power of land certificates Table 3 Evaluation of legal land rights and the corresponding articles in land codes. Kazakh farmers Uzbek farmers Landowners Land tenants Land tenants Rights Value Articles Value Articles Value Articles Access 5 23(2) 5 12(17), 37 (1) 5 17, 39, 46, 49 Withdrawal 5 25(1) 5 64(1/2) 5 39 Land use change 3 65(4), 97 (3–2), 98 3 65(4), 97 (3–2), 98 1 43, 49 Management 5 25(1),42 (2), 64 (1/1) 5 42(2), 64 (1–1) 3 12 Investment 5 65(1/ 2.9), 25 (1) 5 65(1/2,1/ 9) 3 13, 40 Exclusion 5 164 5 33(1), 65 (1/ 8),164–1 (1) 5 53 Income generating 5 64(1/2), 25(1) 5 64(1/2) 5 39 Reallocation 5 25(2) 1 33(2) 1 53 Sell 5 25(2) 1 33(2) 1 53 Renting out 5 25(3) 1 33(2), 38 (5) 1 53 Leasing from land tenants 1 33(2), 38 (5) 1 33(2), 38 (5) 1 52 Inheritance 5 40 5 40 1 17,19 Protection by courts 5 5, 164 5 5, 29(2),32 (1), 35 (4,5), 64 (1.4), 164 5 41, 83, 85, 89 Power of land certificates 5 12(24), 43(10), 97(7), 171 5 12(24), 43 (10), 97(7), 171 5 20, 33, 35 Source: Own assessment based on the Land codes of Kazakhstan (2003) and Uzbekistan (1998) Note: The survey was conducted in 2019; therefore, the assessment does not include amendments in Uzbek land code introduced in 2020. Z. Akhmadiyeva and T. Herzfeld Land Use Policy 109 (2021) 105726 7 are expected to affect the farmers’ decisions; however, the signs of their impacts are ambiguous across microeconomic studies on land use (Ma et al., 2015b; Qu et al., 2018). Farm-specific variables that may affect productivity or cost of cultivating include farm size, distance to the nearest market, soil fertility, salinity, and irrigation conditions. Economies of scale would predict declining costs per ha; however, supervision and other costs might increase with increasing farm size. Distance to the nearest market is expected to have a negative effect on farmers’ decisions, as an increasing remoteness might result in lower farm gate prices, land investment, and input use. The latter three agroecological attributes of farmland provide approximate evaluations by farmers and do not constitute actual physical measures. Land fertility and irrigation conditions are expected to impact positively on the production increase, while soil salinity is expected to impact negatively. Finally, we introduced a dummy variable for Uzbekistan to control for unobserved country differences. In view of such determinants as irrigation infrastructure, water availability, access to production inputs and local organisations, village dummy variables may give more adequate results to capture village-level variation, but this cannot be done due to the lack of the respective data. The list of additional explanatory variables with descriptions is presented in Table 5. 4.3. Model specification and estimation strategy To test the relationship between farmers’ intention to increase production and LR-AP discrepancies, we follow the approach of previous studies (Brasselle et al., 2002; Twerefou et al., 2011). Intention can be expressed as a function of a vector of explanatory variables, Xi, among which institutional indicators are of our interest: I∗ i=β′Xi+ui,ui∼N(0,1)(2) where β′is a vector of unknown parameters. The dependent variable, I∗ i, is latent and consists of the different likelihood levels of production increase, as revealed by the responses, identified by the ith farmer. An observable variable Ii that is a collapsed version of I∗ iis ordinal and depends on various threshold points of I∗ i. Several studies have indicated that land rights are endogenous due to simultaneity or reverse causality from land improvements to tenure security (Besley, 1995; Brasselle et al., 2002; Twerefou et al., 2011). For this reason, it would be crucial to control for omitted variables, which drive perceived security or discrepancies and willingness to gain higher output at the same time. To capture the potential distortion from endogeneity in our estimates, we use the two-step conditional maximum likelihood (2SCML) approach proposed by Rivers and Vuong (1988). The reason for using 2SCML is that the dependent variable and our focus endogenous variables are discrete. The conventional two-stage least squares model would yield biased estimators under these conditions (Brasselle et al., 2002). Initially, the 2SCML procedure was developed for binary probit regression; however, it has proven to deal with ordered probit as well (Dow, 2008). The first stage of 2SCML includes the estimation of a linear probability regression for the discrete endogenous variable by using instrumental variables to generate the estimated vector of residuals. The second stage involves the estimation of ordinal probit maximum likelihood by adding the vector of residuals from the first stage. Coefficients of the first stage residuals can be used to test the endogeneity of corresponding variables (Rivers and Vuong, 1988). Finding appropriate instrumental variables for LR-AP discrepancies is a complicated issue. To overcome this problem, we used a heteroscedasticity-based instrumental method proposed by Lewbel (2012) that allows constructing instruments in the absence of traditional identification. The traditional way to obtain identification is to find instruments satisfying the exclusion restriction that implies no direct effects of the instruments on the dependent variable in the second-stage regression. However, there are cases when no instrument is excluded, or when validation studies are not available. The principle of Lewbel’s 10 is that constructing valid instruments for an endogenous variable can be achieved by exploiting heteroscedasticity in the first stage model of the Lewbel’s. Normally, this approach has four stages, each of which is calculated using an ordinary least squares estimator. Since our dependent variable and potentially endogenous variables are ordinal, we use Table 4 Descriptive statistics of discrepancies. Kazakhstan Uzbekistan Rights Discrepancy Mean Std. Dev. Mean Std. Dev. 1 Access positive 0.11 0.37 0.08 0.43 2 Withdrawal positive 0.29 0.56 2.50 1.54 3 Land use change positive 0.08 0.36 4 negative -1.66 0.63 -0.47 0.93 5 Management positive 0.26 0.54 0.28 0.49 6 negative -0.20 0.45 7 Investment positive 0.33 0.68 0.13 0.49 8 negative -1.43 0.73 9 Exclusion positive 0.47 1.03 0.47 1.15 10 Income generating positive 0.27 0.58 3.43 0.96 11 Reallocation positive 0.67 1.33 12 negative -0.49 1.14 -0.34 0.69 13 Sell positive 0.54 1.26 14 negative -0.55 1.31 0.00 0.07 15 Renting out positive 0.47 1.10 16 negative -0.66 1.36 -0.27 0.54 17 Lease from land tenants negative -2.84 1.48 -0.18 0.50 18 Inheritance positive 0.12 1.18 19 negative -0.14 0.42 20 Protection by courts positive 1.43 0.93 1.93 0.65 21 Power of land certificates positive 0.39 0.85 0.82 0.87 Table 5 Definitions of additional explanatory variables. Variable Description Mean SD Farmer characteristics Educational level Level of last completed education degree: from 1 =no education to 8 =university 5.6 1.7 Age Age of farmer in years 44.5 11.9 Special agricultural education Dummy variable for special education: 1=have; 0 =otherwise 0.3 Consultancy services Dummy variable for using consulting services: 1 =have; 0 =otherwise 0.1 Desired period Number of years a farmer desires to use his land:1 =up to 3 years, 2 =up to 5 years, 3=up to 10 years, 4 =more than 10 years 3.7 0.7 Farmland characteristics Farm size Total land area (ha) 25.4 28.2 Distance to market Average distance between farm and the nearest market 15.1 10.9 Soil fertility Weighted average of soil fertility evaluation: 1 =not good for cultivation, 2=good for 1 crop per year, 3 =good for 2 crops per year, 4 =good for >2 crops per year 2.6 0.6 Salinity Weighted average of soil salinity evaluation: 1 =non-saline, 2 =low saline, 3=medium saline, 4 =high saline. 2.1 1.1 Irrigation conditions Conditions of irrigation and drainage network: 1 =bad, 2 =satisfying, 3 =good 2.1 0.7 Regional characteristics Uzbekistan Dummy variable for Uzbekistan:1 =farmer resides in Uzbekistan; 0 =otherwise 10 Following Rao et al. (2017), we use the term “the Lewbel’s” for a heteroscedasticity-based estimator. Z. Akhmadiyeva and T. Herzfeld Land Use Policy 109 (2021) 105726 8 the first three stages from the Lewbel’s to construct instruments; then we proceed with the second stage of 2SCML. Similar procedures were applied by Rao et al. (2017) with binary probit maximum likelihood in the second stage of 2SCML. In particular, our estimation approach involves the following steps: Step (1): In line with the Lewbel’s procedures, we run the linear probability regression for each of 21 discrepancy variables (Discrepancyj)on a vector of exogenous variables (X)that are control variables in our study: Discrepancyj=β0,j+∑βjX+res1disc,jfor j =1…21 (3) Step (2): To generate instruments, exogenous variables were standardized (XZ)and multiplied with residuals from Eq. (3), (XZ∙res1disc,j). Thus, we obtained 21 sets of constructed instruments for each of the discrepancies. Step (3): We performed the first stage of 2SCML for each of the discrepancies by plugging generated instrumental variables into linear probability regression and computed the respective residuals, res2disc,j: Discrepancyj=β0,j+∑β1,jXZ⋅res1disc,j+∑β2,jX+res2disc,jfor j =1…21 (4) Step (4): Residuals from Eq. (4) were added to the second stage of 2SCML, Eq. (5). Finally, we used the ordered probit model to regress farmers’ intention (Ik)to the vector of original endogenous (Discrepancy) and exogenous (X) variables and the vector of residuals (res2disc)corresponding to each of endogenous variables. Ik=β0k+∑β1,k∙Discrepancy +∑β2,kX+∑β3,k∙res2disc + ε k(5) In Eq. (5), k varies from 1 to 4, because we run four regressions for each of the bundles of the LR-AP discrepancies. Due to the presence of heteroscedasticity, we used robust standard errors to adjust the estimate of the variance-covariance matrix of residuals in the final step. 5. Results and discussion 5.1. Comparison of discrepancies between land rights and actual tenure practices Data on actual practices collected during the survey was used to estimate their discrepancies with legal rights (see Table 1A in the Appendix for descriptive statistics of AP and discrepancies by tenure types). We summarize results in Fig. 2 that demonstrates the shares of null, positive, and negative discrepancies for land users with different sets of land rights. Abbreviations in the figure stand for full legal right (FR), limited legal right (LLR), and no legal right (NR) for each component of the bundle of rights. Kazakh landowners represent the most beneficial group in terms of no-discrepancies. Kazakh land tenants produce the least congruent results among the three groups, although Uzbek farmers have more restricted land rights. The right to land use change and the right to lease land from other farmers who rent state land are of special interest. For both categories of Kazakh farmers, the extremely high share of negative discrepancies, about 90 per cent of the corresponding samples, indicates that farmers practice land transfer which they are formally not allowed to. The land code prohibits changing the designation and the use provision of a certain agricultural plot. This situation indicates a weak enforcement of land law and the efforts of Kazakh farmers to maximize land value by transferring land to more effective users, even at the cost of tenure security for those who initiates the transfer (land can be withdrawn by government). Uzbek farmers, who have limited right to make any investment in land improvement independently, show that this limitation is being often violated. However, most of them uphold prohibitions in land alienation and leasing from other farmers. The interesting point is that Uzbek farmers heavily underuse the right to withdrawal and income generating. The reason for this is that most of the Uzbek respondents are cotton and wheat producers who follow state orders in production and mandatory sales plan. All three categories have a similar pattern for discrepancies in the government protection bundle. Most of the farmers perceive less government protection of their rights in courts than postulated by law. To assess the actual practice of government protection, we estimate the Fig. 2. Shares of positive, negative and no discrepancies between legal rights and actual practices by tenant types. Note: The sample size of Kazakh farmers with private land is 331; of Kazakh farmers with rented land is 164, of Uzbek farmers is 460. Abbreviations stand for FR=full right, LLR=limited right, and NR=no right for each of land rights. Z. Akhmadiyeva and T. Herzfeld